Search

The enhancement of low-quality images is both a challenging task and an essential endeavor in many fields including computer vision, computational photography, and image processing. In this paper, we propose a novel and fully explainable method for image enhancement that combines spatial selection and histogram equalization. Our approach leverages tree-search theory and deep reinforcement learning to iteratively select areas to be processed. Extensive experimentation on two datasets demonstrates the quality of our method compared to other state-of-the-art models. We also conducted a multi-user experiment which shows that our method can emulate a variety of enhancement styles. These results highlight the effectiveness and versatility of the proposed method in producing high-quality images through an explainable enhancement process. • A fully explainable image enhancement method based on reinforcement learning. • The method alternates spatial selection and histogram equalization through deep RL. • An extensive experimentation shows that our method is competitive with SOTA methods. [ABSTRACT FROM AUTHOR]

Recently, computer vision and artificial intelligence are being used as enabling technologies for plant phenotyping studies, since they allow the analysis of large amounts of data gathered by the sensors. Plant phenotyping studies can be devoted to the evaluation of complex plant traits either on the aerial part of the plant as well as on the underground part, to extract meaningful information about the growth, development, tolerance, or resistance of the plant itself. All plant traits should be evaluated automatically and quantitatively measured in a non-destructive way. This paper describes a novel approach for identifying plant roots from images of the root system architecture using a convolutional neural network (CNN) that operates on small image patches calculating the probability that the center point of the patch is a root pixel. The underlying idea is that the CNN model should embed as much information as possible about the variability of the patches that can show chaotic and heterogeneous backgrounds. Results on a real dataset demonstrate the feasibility of the proposed approach, as it overcomes the current state of the art. • Root systems must be monitored to assess the growth and well-being of a plant. • State-of-the-art approaches mainly use classic ML or U-networks for segmentation. • CNNs can be used for monitoring considering patch-based information. • These models are simpler and faster, and provide better segmentation performance. [ABSTRACT FROM AUTHOR]

Deep fake technology paves the way for a new generation of super realistic artificial content. While this opens the door to extraordinary new applications, the malicious use of deepfakes allows for far more realistic disinformation attacks than ever before. In this paper, we start from the intuition that generating fake content introduces possible inconsistencies in the depth of the generated images. This extra information provides valuable spatial and semantic cues that can reveal inconsistencies facial generative methods introduce. To test this idea, we evaluate different strategies for integrating depth information into an RGB detector and we propose an attention mechanism that makes it possible to integrate information from depth effectively. In addition to being more accurate than an RGB model, our Masked Depthfake Network method is + 3. 2 % more robust against common adversarial attacks on average than a typical RGB detector. Furthermore, we show how this technique allows the model to learn more discriminative features than RGB alone. • Integrating depth and RGB improves the accuracy and robustness of deepfake detectors. • Late fusion is the best fusion strategy for integrating RGB and depth features. • We guide the integration of depth information via a self-attention mechanism. • Depth integration makes the model more robust to some opposing attacks. [ABSTRACT FROM AUTHOR]

Active vision aims to equip computer vision methods with the ability to dynamically adjust the capturing sensor's viewpoint, position, or parameters in real time. This dynamic capability allows for improving the accuracy of the perception process. However, training and evaluating an active vision model often requires a large number of annotated images captured under different sensor and environmental settings, in order to emulate actions like moving around, approaching, or moving away from a person and thus effectively model the active perception dynamics. Obviously, collecting and annotating such datasets is a challenging and expensive task. To overcome these limitations, this paper introduces a synthetic image generation pipeline specifically designed to support active vision tasks. The pipeline is developed using a highly realistic simulation framework based on Unity and allows for the generation of images depicting humans, captured at varying view angles, distances, illumination conditions, and backgrounds, supporting a wide range of different tasks. Two annotated datasets, namely ActiveHuman and ActiveFace, are generated using the pipeline and the effectiveness of the proposed approach is demonstrated by a solid use case that involves training and evaluating an embedding-based active face recognizer. Furthermore, we demonstrate how the proposed generation approach enables expanding existing active face recognition methods by training models that control both the left/right movements, as well as the distance to a subject, leveraging the additional information provided by ActiveFace dataset. To facilitate replication and encourage the use of the generated datasets for training and evaluating other active vision approaches, the associated assets and the developed dataset generation pipeline is to become publicly available. • Active vision models require a large number of annotated data samples. • We propose a synthetic image generation pipeline designed to support active vision. • A highly realistic simulation framework based on Unity is used. • Two annotated datasets (ActiveHuman and ActiveFace) are generated. • Evaluation provided on a challenging active face recognition setup. [ABSTRACT FROM AUTHOR]

Hyperspectral image (HSI) classification is an active research problem in computer vision and multimedia field. Contrary to traditional image data, HSIs contain rich spectral, spatial and semantic information. Thus, how to extract the discriminative features for HSIs by integrating spectral , spatial and semantic cues together is the core issue to address HSI classification task. Existing works mainly focus on exploring spectral and spatial information which usually fail to fully explore the rich semantic information in HSIs. To address this issue, in this paper, we first propose a novel semantic Transformer scheme, named SemanticFormer, which aims to learn discriminative visual representations for semantics by exploiting the interaction among different semantic tokens. Using the proposed SemanticFormer, we then propose a novel heterogeneous network that contains both spectral–spatial convolution network branch and SemanticFormer branch to extract spectral–spatial and semantic features simultaneously for HSIs. Experiments on two widely used datasets demonstrate the effectiveness of our SemanticFormer and HSI classification network method. Our codes will be available in https://github.com/SissiW/SemanticFormer. • We propose a novel SemanticFormer module for semantic representation. • We integrate SemanticFormer into HSI classification framework. • SemanticFormer is a general scheme in CV. • Extensive experimental results prove the effectiveness of SemanticFormer. [ABSTRACT FROM AUTHOR]

This paper explores the representation of vehicle lights in computer vision and its implications for various pattern recognition tasks in autonomous driving. Different representations for vehicle lights, including bounding boxes, center points, corner points, and segmentation masks, are discussed in terms of their strengths and weaknesses toward a variety of domain tasks, as well as associated data collection and annotation challenges. This leads to the introduction of the LISA Vehicle Lights Dataset, providing light annotations related to position, state, color, and signal, specifically designed for downstream applications in vehicle detection, intent and trajectory prediction, and safe path planning. A comparison of existing vehicle light datasets is provided, highlighting the unique features and limitations of each dataset. Because occlusions from vehicle pose and passing objects can limit camera observation, we introduce a group of Light Visibility neural networks, which take as input a detected vehicle image and return as output whether a corresponding vehicle light is present in the image. This is especially important for the evaluation of light localizations, states, and signals, as systems decisions should account for differences of lights being in unknown states due to occlusion versus model uncertainty. We show that our trained Light Visibility Models achieve over 90% accuracy on each of the four light classes. Our dataset and model are made available at https://cvrr.ucsd.edu/vehicle-lights-dataset. • Discussion of tradeoffs in representations of vehicle lights for data-driven models. • Introduction of the LISA Vehicle Lights Dataset. • Light visibility classification results for downstream tasks in pattern recognition. • Description of pattern recognition tasks which rely on light detection. • Discussion of importance of association of lights to particular vehicle instances. [ABSTRACT FROM AUTHOR]

• Novel CNN model is proposed for efficiently solving jigsaw puzzle as pretext task. • The model is trained to learn explainable visual representational features. • Proposed Divide-and-Teach strategy helps in resource-constrained training. • It is effective to extract explainable and transferable context-invariant features. • First work on self-supervised learning for ACL tear detection in Knee MR. The success of deep learning based models for computer vision applications requires large scale human annotated data which are often expensive to generate. Self-supervised learning, a subset of unsupervised learning, handles this problem by learning meaningful features from unlabeled image or video data. In this paper, we propose a self-supervised learning approach to learn transferable features from MR video clips by enforcing the model to learn anatomical features. The pretext task models are designed to predict the correct ordering of the jumbled image patches that the MR video frames are divided into. To the best of our knowledge, none of the supervised learning models performing injury classification task from MR video provide any explanation for the decisions made by the models and hence makes our work the first of its kind on MR video data. Experiments on the pretext task show that this proposed approach enables the model to learn spatial context invariant features which help for reliable and explainable performance in downstream tasks like classification of Anterior Cruciate Ligament tear injury from knee MR videos. The efficiency of the novel Convolutional Neural Network proposed in this paper is reflected in the experimental results obtained in the downstream task. The proposed model achieves an accuracy of 76.62% and an AUC score of 0.848 on the Sagittal plane, outperforming the contrastive learning algorithms like PIRL and SimCLR using jigsaw puzzle as a transformation. The proposed model also achieved an AUC score of 0.740 on the KneeMRI dataset. [ABSTRACT FROM AUTHOR]

Given the visual data of an object, directly predicting the high-degree-of-freedom grasping pose of a dexterous hand is a challenging task. In this paper, we propose a new grasp synthesis framework based on hand-object interaction for representing the grasping pose of a high degree of freedom hand. First, the attention operation embedded in the finger encoder guides the contact of each finger with the object to obtain the initial grasp pose; then, based on the hand-object interaction constraints, the pose is further optimized in the finger refinement module. The experimental results show that our model enables the robot to grasp objects stably. • A new grasp synthesis framework for dexterous hands based on hand-object interaction. • Cascaded finger joint angle estimation strategy to model hand-object interaction. • A refinement module with hand-object contact constraints to improve the grasp quality. • High grasping success rate for most experimental objects in multiple categories. [ABSTRACT FROM AUTHOR]

Text-to-image (T2I) generation, which involves synthesizing an image from a textual description, has emerged as a popular research topic in computer vision. Meanwhile, transformer-based models, such as BERT, GPT-2, and T5, have demonstrated promising results in various natural language processing tasks, including text generation and translation. However, their application to T2I generation remains largely unexplored. Therefore, a comparative study investigating the performance of BERT, GPT-2, and T5 in T2I generation is of significant importance. Such a study would shed light on the strengths and weaknesses of each model, facilitating the identification of the most suitable approach for this task. In this paper, we propose three architectures to conduct a comparative study of T5, GPT-2, and BERT in T2I generation tasks. We fine-tune these models to generate text vectors and then transform the textual information into images using the affine transformation into the DF-GAN generator. Subsequently, we evaluate the quality, diversity, and the models' ability to recognize the words. Our experiments on the challenging CUB and Oxford-102 flower datasets demonstrate that T5 exhibits promising potential for T2I generation. It has the capability to generate visually appealing and semantically coherent images from textual descriptions. • GPT-2, BERT, and T5 have succeeded in synthesizing the images from the textual description with varying degrees of success. • BERT allows the model to capture the relationships between words in the input text and generate a more coherent image. • GPT-2 allows the model to generate plausible images, but the quality of the images is often lower compared to other models. • T5 allows the model to be more flexible, it is the powerful transformer-based model for text-to-image generation tasks. • The quality of the generated images is still lower compared to other generation models. [ABSTRACT FROM AUTHOR]

• The paper presents concrete applications of Visual Question Answering • Domains where VQA has been experimented are presented together with the exploited dataset • The paper suggests some challenging techniques that can be especially suited for specific domains • Some final considerations sketch future work in domain-related VQA Visual Question Answering (VQA) is an extremely stimulating and challenging research area where Computer Vision (CV) and Natural Language Processig (NLP) have recently met. In image captioning and video summarization, the semantic information is completely contained in still images or video dynamics, and it has only to be mined and expressed in a human-consistent way. Differently from this, in VQA semantic information in the same media must be compared with the semantics implied by a question expressed in natural language, doubling the artificial intelligence-related effort. Some recent surveys about VQA approaches have focused on methods underlying either the image-related processing or the verbal-related one, or on the way to consistently fuse the conveyed information. Possible applications are only suggested, and, in fact, most cited works rely on general-purpose datasets that are used to assess the building blocks of a VQA system. This paper rather considers the proposals that focus on real-world applications, possibly using as benchmarks suitable data bound to the application domain. The paper also reports about some recent challenges in VQA research. [ABSTRACT FROM AUTHOR]

• This work proposes a simple loss function for imbalanced image classification. • This work studies the effect of suppressing output scores on incorrect classes for imbalanced image classification. • This work demonstrates the effectiveness of the proposed method through experiments on imbalanced datasets. [Display omitted] Recently, deep learning models have achieved great success in computer vision applications, relying on large-scale class-balanced datasets. However, imbalanced class distributions still limit the wide applicability of these models due to degradation in performance. To solve this problem, in this paper, we concentrate on the study of cross entropy which mostly ignores output scores on incorrect classes. This work discovers that neutralizing predicted probabilities on incorrect classes improves the prediction accuracy for imbalanced image classification. This paper proposes a simple but effective loss named complement cross entropy based on this finding. The proposed loss makes the ground truth class overwhelm the other classes in terms of softmax probability, by neutralizing probabilities of incorrect classes, without additional training procedures. Along with it, this loss facilitates the models to learn key information especially from samples on minority classes. It ensures more accurate and robust classification results on imbalanced distributions. Extensive experiments on imbalanced datasets demonstrate the effectiveness of the proposed method compared to other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

• Temporal convolutional network with temporal position fusion for estimating evoked expressions in video. • Address the inconsistency of time steps during training and testing with temporal position fusion. • State-of-the-art result on Evoked Expressions from Videos (EEV) dataset. • Made huge improvements with temporal position fusion on MediaEval 2018 dataset. This paper introduces an approach for estimating evoked categories expression from videos with the temporal position fusion. Pre-trained models on large-scale datasets in computer vision and audio signals were used to extract the deep representation for timestamps in the video. A temporal convolution network, rather than an RNN-like architecture, was applied to explore temporal relationships due to its advantage in memory consumption and parallelism. Furthermore, to address the noise labels, the temporal position was fused with the deep learned feature to ensure the network differentiates the time steps when noise labels were removed from the training set. This technique helps the system gain a considerable improvement compared to other methods. We conducted experiments on EEV, a large-scale dataset for evoked expression from videos, and achieved a score of 0.054 in terms of Pearson correlation coefficient as a state-of-the-art result. Further experiments on a sub set of LIRIS-ACCEDE dataset - MediaEval 2018 benchmark, also demonstrated the effectiveness of our approach. [ABSTRACT FROM AUTHOR]

• Substituting the FFNs of the transformer encoders with locality modules enriches the local contexts in multi-object tracking. • The quantitative and qualitative research shows the strength of locality in transformer-based MOT systems. • The performance of pedestrian tracking is boosted when training the model with a mixture of detection and tracking datasets. Multi-object tracking (MOT) is a critical task in various domains, such as traffic analysis, surveillance, and autonomous vehicles. The joint-detection-and-tracking paradigm has been extensively researched, which is faster and more convenient for training and deploying over the classic tracking-by-detection paradigm while achieving state-of-the-art performance. This paper explores the possibilities of enhancing the MOT system by leveraging the prevailing convolutional neural network (CNN) and a novel vision transformer technique Locality. There are several deficiencies in the transformer adopted for computer vision tasks. While the transformers are good at modeling global information for a long embedding, the locality mechanism, which learns the local features, is missing. This could lead to negligence of small objects, which may cause security issues. We combine the TransTrack MOT system with the locality mechanism inspired by LocalViT and find that the locality-enhanced system outperforms the baseline TransTrack by 5.3% MOTA on the MOT17 dataset. [ABSTRACT FROM AUTHOR]

• Naïve transfer techniques lead to feature space collapse in anomaly detection and segmentation area. • Topology-preserve constraint can mitigate collapse conspicuously. • Significant improvement over previous methods on BTAD and CIFAR100/10 datasets. Models pre-trained on the ImageNet dataset are introduced to be exploited for knowledge transfer in numerous downstream computer vision tasks, including the weakly-supervised anomaly detection and segmentation area. Specifically, in anomaly segmentation, the former study shows that representing images with feature maps extracted by pre-trained models significantly improves over previous techniques. This kind of representation method requires both high-quality and task-specific features, but feature extractors obtained from ImageNet directly are very general. One intuition for obtaining stronger features is by transferring a pre-trained model to the target dataset. However, in this paper, we show that under weakly-supervised settings, naïve fine-tune techniques that typically work for supervised learning can lead to catastrophic feature space collapse and reduce performance greatly. Thus, we propose to apply a topology-preserving constraint during transferring. Our method preserves the topology graph to keep the feature space from collapsing under weakly-supervised settings. And then we combine the transferred model with a simple anomaly detection and segmentation baseline for performance evaluation. The experiments show that our method achieves competitive accuracy on several benchmarks meanwhile setting a new state-of-the-art for anomaly detection on CIFAR100/10 and BTAD datasets. [ABSTRACT FROM AUTHOR]

• A crack data set with various forms is constructed by manual collection. • Proposes a network model suitable for multi shape object segmentation. • Proposes a kind of boundary weighted loss function which is suitable for the segmentation network of small objects. • Our model has better detection effect than the existing crack segmentation network in this scene. The existence of cracks will affect the stability of the building. It is very important to identify and deal with the cracks in time to ensure the safety and stability of the building. Based on the above background, the purpose of this paper is to study the method of building crack recognition and total quality management based on deep learning. This paper focuses on the computer vision technology in artificial intelligence, studies the image classification algorithm and semantic segmentation algorithm based on the deep learning method, and applies it to the field of building crack image analysis. In this paper, we use the deep convolution neural network to design the building image crack classification model and segmentation model, realize the identification and analysis of building cracks, and build a building crack analysis system, which can significantly improve the efficiency of building crack detection. Then, based on the image processing technology, the quantitative analysis of the fracture segmentation results is carried out. Through the basic morphological methods such as corrosion, expansion, opening and closing operations, the segmentation mark map, skeleton map and geometric parameter information of the fracture are obtained, which further provides the maintenance and judgment basis for professional engineers. The experimental results show that compared with FCN, the accuracy of rfcn-a is improved by 5.98%, the precision is improved by 6.07%, and the real and f'score are improved by 3.11% and 6.01%, respectively. [ABSTRACT FROM AUTHOR]

• The two change detection algorithms proposed in this paper use KFCM gradient clustering algorithm. • KFCM algorithm is a classic clustering algorithm in unsupervised clustering, which can effectively cluster most of the data. • We will study how to use the special features of the image to complete the clustering of the difference maps. In recent years, computer vision, especially deep learning, has been widely used in various fields. Through the deep learning aerial image detection gradient clustering algorithm automatic recognition, it can solve the limitations of manual shooting by humans, can shoot from a high altitude to a panoramic view of a specific area, and provide a more comprehensive solution. The traditional forest resource management and management work is mainly carried out by forestry personnel to carry out a large number of investigations and investigations on the forest. This method not only consumes a lot of manpower and material resources, but also does not have real-time nature. It is difficult to deal with all kinds of forest management. Problems, causing unnecessary losses. In this regard, this paper proposes an aerial image change detection algorithm based on H-KFCM, and designs related experiments to verify and demonstrate the performance of the algorithm. In this paper, we conduct a parallel study based on deep learning on the gradient clustering algorithm of deep learning in aerial image processing. By using CUDA (Compute Unified Device Architecture) to perform large-scale parallel processing of aerial data. Can greatly shorten the time to obtain results, improve the efficiency of relevant personnel. Experiment analysis. It can be seen from the results that the deep learning parallelization program implemented in this paper has a faster calculation speed and uses less time in high-resolution images, and has a good acceleration ratio compared to the CPU. [ABSTRACT FROM AUTHOR]

• We devise the entropy regularization method as a new training strategy for WSOL. • The proposed regularization method is efficient in terms of computational overheads. • Our method is very easy to implement, requiring only one line of code. • We demonstrate the effectiveness of our method on five different datasets. The goal of weakly-supervised object localization (WSOL) is to train a localization model without the location information of the object(s). Recently, most existing WSOL methods capture the object with an attention map extracted from a classification network. However, it has been observed that we need to sacrifice classification performances to achieve the best WSOL score. We conjecture that this is because the objective of classification training, minimizing entropy between one-hot ground truth and predicted class probability, is not entirely consistent with that of localization. In this paper, we investigate how the entropy of predicted class probability affects localization performances, where we conclude that there is a sweet spot for localization with respect to entropy. Hence, we propose a new training strategy that adopts entropy regularization for finding the optimal point effectively. Specifically, we add an additional term to the loss function, which minimizes the entropy between a uniform distribution and the predicted class probability vector. The proposed method is easy to implement since we do not need to modify the architecture or data pipeline. In addition, our method is efficient in that almost zero additional resources are required. Most importantly, our method improves WSOL scores significantly, which has been shown through extensive experiments. [ABSTRACT FROM AUTHOR]

• A multi-scale local transformer is proposed for modeling the correlation between frames in various-range local windows. • A deep network is combined by the 3D feature extractor, local transformer and multi-head estimator. • The promising results on two publically-available datasets are achieved for spotting micro-expressions. Micro-expression analysis by computer vision techniques has attracted much attention as it can reveal the human emotions automatically. Among the analysis tasks, the temporal spotting is the most challenging task for achieving expression-aware frames from long video sequences. Compared to the well studied recognition task, more researches need to be devoted to the spotting task for further improving the performance and benefiting the subsequent tasks. So, in this paper, we propose a convolutional transformer based deep model for micro-expression spotting in long video sequences. A 3D convolutional subnetwork is firstly employed to extract the visual features from the temporal frames in a fixed-size sliding window of original video sequence. Then a multi-scale local transformer module is designed based on the visual features to model the correlation between frames in a local window. By leveraging the correlation information, the description of face movement becomes more representative for various-duration micro-expressions. Finally, the multi-head classifier and the corresponding estimator are jointly combined to predict the temporal position for spotting micro-expressions. The proposed method is evaluated on two publicly-available datasets, namely CAS(ME) 2 and SAMM-LV, and achieves the promising performance of 0.2770 F1-score on SAMM-LV and 0.1373 F1-score on CAS(ME) 2. The code is publicly available on GitHub (https://github.com/xiazhaoqiang/MULT-MicroExpressionSpot). [ABSTRACT FROM AUTHOR]

• Creation of a new face detection dataset (MAFD-150) based on modern art. • Describes face detection on modern art as a challenging task for Visual Turing test. • Analysis of the characteristics of modern art that make it challenging for algorithms. • Challenge to claims of "human-level performance" on face detection and recognition. • Performance evaluation for face detection on modern art. There is a widely held belief that computer vision, in general, and face authentication, in particular, are to a large extent solved problems. This paper challenges this belief regarding face authentication using examples from modern art that significantly confound face detection. The challenges are made concrete using a new MAFD-150 dataset (M odern A rt F ace D etection) composed mostly of modern art examples that cover much diversity in style and artists. MAFD-150 challenges the belief that singleton and crowd face detection is an almost solved problem, and provides baselines and preliminary results that highlight the inadequacy of current expertise and methods to address face detection. In particular, we show that well-known face detection algorithms are only able to achieve an F1 score of less than 35% overall across the new dataset. Additionally, we discuss the performance of the selected face detectors on varying art categories (such as Impressionism, Pop Art, et al.) to show how style and face representation may impact these algorithms. The paper concludes with suggestions on how to advance face processing by leveraging the complementarity between Show-and-Tell-like methods and a context and cooperative driven visual question answering framework using relevance-based triage. The very challenges detailed throughout are then shown to be helpful with developing novel, robust, and secure access protocols that combine text and modern art images using the visual question answering framework. [ABSTRACT FROM AUTHOR]

• We propose a Transformer-based network architecture for video salient object detection. • We propose a Gated Cross Reference (GCR) module to facilitate collaborative learning between appearance and motion cues. • The proposed Transformer-based method reaches excellent performance on five public benchmarks. Video salient object detection is a fundamental computer vision task aimed at highlighting the most conspicuous objects in a video sequence. There are two key challenges presented in video salient object detection: (1) how to extract effective feature representations from appearance and motion cues, and (2) how to combine both of them into robust saliency representation. To handle these challenges, in this paper, we propose a novel Transformer-based Cross Reference Network (TCRN), which fully exploits long-range context dependencies in both feature representation extraction and cross-modal (i.e., appearance and motion) integration. In contrast to existing CNN-based methods, our approach formulates video salient object detection as a sequence-to-sequence prediction task. In the proposed approach, the deep feature extraction is achieved by a pure vision transformer with multi-resolution token representations. Specifically, we design a Gated Cross Reference (GCR) module to effectively integrate appearance and motion into saliency representation. The GCR first propagates global context information between different modalities, and then perform cross-modal fusion by a gate mechanism. Extensive evaluations on five widely-used benchmarks show that the proposed Transformer-based method performs favorably against the existing state-of-the-art methods [ABSTRACT FROM AUTHOR]

• A diverse captioning model of full convolution design is proposed. • We develop a new evaluation metric to assess the sentence diversity. • Our method achieves superior performance compared to state-of-the-art benchmarks. Automatically describing video content with text description is challenging but important task, which has been attracting a lot of attention in computer vision community. Previous works mainly strive for the accuracy of the generated sentences, while ignoring the sentences diversity, which is inconsistent with human behavior. In this paper, we aim to caption each video with multiple descriptions and propose a novel framework. Concretely, for a given video, the intermediate latent variables of conventional encode-decode process are utilized as input to the conditional generative adversarial network (CGAN) with the purpose of generating diverse sentences. We adopt different Convolutional Neural Networks (CNNs) as our generator that produces descriptions conditioned on latent variables and discriminator that assesses the quality of generated sentences. Simultaneously, a novel DCE metric is designed to assess the diverse captions. We evaluate our method on the benchmark datasets, where it demonstrates its ability to generate diverse descriptions and achieves superior results against other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

• Create highly distinguishable similarity scores by a bidirectional dropout method. • Handle task adaptive open-set recognition by a non-parametric Bayesian based metric. • Improve the confidence of similarity scores by a hierarchical cross entropy loss. • Experiments demonstrate our method achieves high performance and generalizability. Existing deep learning approaches have achieved high performance in encrypted network traffic analysis tasks. However, practical requirements such as open-set recognition on dynamically changing tasks (e.g., changes in the target website list), challenge existing methods. While few-shot learning and open-set recognition methods have been proposed for domains such as computer vision, few-shot open-set recognition for encrypted network traffic remains an unexplored area. This paper proposes a task adaptive siamese neural network for open-set recognition of encrypted network traffic with bidirectional dropout data augmentation. Our contributions are three-fold: First, we introduce generated positive and negative pairs into the siamese neural network training process to shape a more precise similarity boundary through bidirectional dropout data augmentation. Second, we utilize Dirichlet Process Gaussian Mixture Model (DPGMM) distribution to fit the similarity scores of the negative pairs constructed by the support set of each query task, and create a new open-set recognition metric. Third, by leveraging the extracted features at coarse and fine granular levels, we construct a hierarchical cross entropy loss to improve the confidence of the similarity score. Extensive experiments on a network traffic dataset and the Omniglot dataset demonstrate the superiority and generalizability of our proposed approach. [ABSTRACT FROM AUTHOR]

• A Gated-CNN-BGRU optical model for offline Handwritten Text Recognition systems. • An optical model capable of training with low data volume, for example 49 images. • Benchmark in 5 datasets and partitions arrangement to create 50 data volume scenarios. • The proposed optical model surpasses the state-of-the-art models in the proposed scenarios. Handwritten Text Recognition (HTR) systems have gained interest in fields of academic research and commercial applications. Deep learning techniques, and more precisely Convolutional Neural Networks (CNNs), have enabled many recent successes in the computer vision community. However, due to high computational costs, applying CNNs to many real applications is challenging since the specific training data is restricted in many cases. Therefore, in this paper, we present a Gated-CNN-BGRU optical model capable of dealing with this complex challenge. The proposed model was evaluated on five well-known datasets in HTR (Bentham, IAM, RIMES, Saint Gall, and Washington). Additionally, we redefine the training and validation partitions for each dataset, progressively varying the percentage of data between both partitions to create a total of 50 scenarios with different data volumes. The experiment validates and shows that the proposed model presents statistically significant results, surpassing the current models by an average of 2.96 and 8.91 percentage points in character and word recognition accuracy. In the most complex scenario of using 49 images for training, we achieved character and word precision of 87.25% and 71.54% respectively. That means an improvement of 78.32 and 53.54 percentage points, respectively, of the state-of-the-art optical models. [ABSTRACT FROM AUTHOR]

• Harris Z + updates the HarrisZ detector accounting for modern image matching pipeline design. • Corners extracted by Harris Z + are well-localized and well-distributed over the image. • Image matching pipelines based on Harris Z + are very competitive on recent benchmarks. Due to its role in many computer vision tasks, image matching has been subjected to an active investigation by researchers, which has lead to better and more discriminant feature descriptors and to more robust matching strategies, also thanks to the advent of the deep learning and the increased computational power of the modern hardware. Despite of these achievements, the keypoint extraction process at the base of the image matching pipeline has not seen equivalent progresses. This paper presents HarrisZ+, an upgrade to the HarrisZ corner detector, optimized to synergically take advance of the recent improvements of the other steps of the image matching pipeline. HarrisZ+ does not only consists of a tuning of the setup parameters, but introduces further refinements to the selection criteria delineated by HarrisZ, so providing more, yet discriminative, keypoints, which are better distributed on the image and with higher localization accuracy. The image matching pipeline including HarrisZ+, together with the other modern components, obtained in different recent matching benchmarks state-of-the-art results among the classic image matching pipelines. These results are quite close to those obtained by the more recent fully deep end-to-end trainable approaches and show that there is still a proper margin of improvement that can be granted by the research in classic image matching methods. [ABSTRACT FROM AUTHOR]

• Rain droplets removed by generating the Laplacian pyramid levels of rain-free images. • The Laplacian pyramid decomposition improves with respect to a simple encoder-decoder. • The training benefits from comparing each generated level with the ground truth. • Reconstructing levels before comparison, balances their impact and provides context. • Better high frequencies will improve structural similarity and signal-to-noise ratio. [Display omitted] Digital raindrop removal is a branch of image restoration that aims at identifying adherent droplets on a glass surface and replacing them with plausible content. When successfully performed, raindrop removal was proven in the past to positively affect both the perceived appearance of the scene, and the performance of computer vision tasks such as semantic segmentation and object detection. In this paper, we design and implement a new encoder-decoder neural network for supervised raindrop removal. Our network, given a rainy input image, produces as output the Laplacian pyramid of a rain-free version of the input, making it possible to handle the variety of appearances of rain droplets by processing different frequency bands independently. To this end, we define and experimentally prove the effectiveness of a custom loss function that combines the errors of the different Laplacian frequency bands. We test our model for raindrop removal on a standard dataset, using multiple objective metrics to provide a detailed analysis of its performance. We confirm the superiority of our proposal in a comparison with other methods from the state of the art. [ABSTRACT FROM AUTHOR]

Weighted Kappa is an index of reference used in many diagnosis systems to compare the agreement between different raters. This index can be also used to evaluate the performance of automatic classification methods against the gold standard given by an expert (or from a consensus of an expert group). On the other hand, in the last years, deep learning has achieved a great importance as a new machine learning method. The usual loss function used in deep learning for multi-class classification is the logarithmic loss. In this paper we explore the direct use of a weighted kappa loss function for multi-class classification of ordinal data, also known as ordinal regression. Three classification problems are solved in the paper using these two loss functions. Results confirm that better classification is made when the model is constructed with the optimization of kappa instead of logarithmic loss. [ABSTRACT FROM AUTHOR]

Estimating parameters of a geometrical model from 3-D point cloud data is an important problem in computer vision. Random sample consensus (RANSAC) and its variations have been proposed for the estimation of the models parameters. However, RANSAC is computationally expensive and the problem is challenging when the measured 3-D data contain noise and outliers. This paper presents an efficient sampling technique for RANSAC, in which geometrical constraints are utilized for selecting good samples for a robust estimation. The constraints are based on two predefined criteria. First, the samples must ensure being consistent with the estimated model; second, the selected samples must satisfy explicit geometrical constraints of the interested objects. The proposed approach is wrapped as a robust estimator, named GCSAC (Geometrical Constraint SAmple Consensus), for estimating a cylindrical object from a 3-D point cloud. Extensive experiments on various data sets show that our method outperforms other robust estimators (e.g. MLESAC) tested in term of both precision of the estimated model and computational time. The implementations and evaluation datasets used in this paper are made publicly available. [ABSTRACT FROM AUTHOR]

• We propose an efficient hybrid ReID backbone for discriminative feature extraction. • We formulate an efficient PW-MSA block to free positions from the fixed length. • We put forward a novel attention-guided GCN model to encode person attributes and body parts into embedding representations. [Display omitted] The Transformer has been applied into computer vision to explore long-range dependencies with multi-head self-attention strategy, therefore numerous Transformer-based methods for person re-identification (ReID) are designed for extracting effective as well as robust representation. However, the memory and computational complexity of scaled dot-product attention in Transformer cost vast overheads. To overcome these limitations, this paper presents ResT-ReID method, which designs a hybrid backbone Res-Transformer based on ResNet-50 and Transformer block for effective identify information. Specifically, we use global self-attention in place of depth-wise convolution in the fourth layer's residual bottleneck of ResNet-50. For fully exploiting the entire knowledge of the person, we devise attention-guided Graph Convolution Networks (GCNs) with side information embedding (SIE-AGCN), which has an attention layer located into two GCN layers. The quantified experiments on two large-scale ReID benchmarks demonstrate that the proposed ResT-ReID achieves competitive results compared with several state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

• The properties of total variation spectral transform are analyzed on the images without and with noise. • An image segmentation framework based on the TV spectral transform is presented. • High segmentation accuracy is shown in the proposed framework compared with existing state-of-the-art methods. • Robustness to noise can be reflected in the proposed framework. Image segmentation is a critical process that influences subsequent tasks in computer vision. Despite years of research, popular segmentation methods usually cause over-segmentation or under-segmentation when segmenting images with noises. To address the issue, this paper proposes an image segmentation framework with the help of total variation (TV) spectral transform technique. Firstly, we impose the TV spectral transform on an image with noises to get basic elements that preserve shape structure. Secondly, the object structure is roughly separated from the background by a separation surface, which is fitted using the maximal response time of each pixel. Thirdly, the roughly generated object structure is refined with a guided filter whose guidance image is calculated through a band-pass filter and inverse transform in the TV spectral domain. Finally, binary segmentation and morphological operations are conducted on the refined object structure to obtain the segmentation result. Experiment results indicate that our method can achieve high segmentation accuracy and be robust to noise compared with other classical approaches. [ABSTRACT FROM AUTHOR]

• A multi-style framework for unsupervised image-to-illustration translation. • A multimodal model: it learns multiple diverse types of styles just by one model. • Our framework uses both local and global information. • It is both input specific and uses the information of other images. • MISS GAN shows a large diversity in its generated images. Unsupervised style transfer that supports diverse input styles using only one trained generator is a challenging and interesting task in computer vision. This paper proposes a Multi-IlluStrator Style Generative Adversarial Network (MISS GAN) that is a multi-style framework for unsupervised image-to-illustration translation, which can generate styled yet content preserving images. The illustrations dataset is a challenging one since it is comprised of illustrations of seven different illustrators, hence contains diverse styles. Existing methods require to train several generators (as the number of illustrators) to handle the different illustrators' styles, which limits their practical usage, or require to train an image specific network, which ignores the style information provided in other images of the illustrator. MISS GAN is both input image specific and uses the information of other images using only one trained model. [ABSTRACT FROM AUTHOR]

• Design a novel Unsupervised DL model for COVID-19 Diagnosis and Classification • Propose a Parameter Tuned Adagrad with Inception v4 model as feature extractor • Employ an Unsupervised Variational Autoencoder Model for classification process • Validate the classification performance on COVID Chest X-ray dataset At present times, COVID-19 has become a global illness and infected people has increased exponentially and it is difficult to control due to the non-availability of large quantity of testing kits. Artificial intelligence (AI) techniques including machine learning (ML), deep learning (DL), and computer vision (CV) approaches find useful for the recognition, analysis, and prediction of COVID-19. Several ML and DL techniques are trained to resolve the supervised learning issue. At the same time, the potential measure of the unsupervised learning technique is quite high. Therefore, unsupervised learning techniques can be designed in the existing DL models for proficient COVID-19 prediction. In this view, this paper introduces a novel unsupervised DL based variational autoencoder (UDL-VAE) model for COVID-19 detection and classification. The UDL-VAE model involved adaptive Wiener filtering (AWF) based preprocessing technique to enhance the image quality. Besides, Inception v4 with Adagrad technique is employed as a feature extractor and unsupervised VAE model is applied for the classification process. In order to verify the superior diagnostic performance of the UDL-VAE model, a set of experimentation was carried out to highlight the effective outcome of the UDL-VAE model. The obtained experimental values showcased the effectual results of the UDL-VAE model with the higher accuracy of 0.987 and 0.992 on the binary and multiple classes respectively. [ABSTRACT FROM AUTHOR]

• A novel nonlinear feature selective networks is designed. • Proposed method is applied for continuous values regression. • An optimizer joins back propagation and re-weighted is designed. The curse of dimensionality is a long-standing intractable issue in many machine learning and computer vision tasks. Feature selection, a data preprocessing technique, aims to select most discriminative feature subsets for improving performance of downstream machine learning tasks. However, most existing feature selection methods concentrate upon learning a linear relationship between data points and their labels, which causes that they are incapable of handling nonlinear complex data in real-world applications. In this paper, we first propose a novel end-to-end Nonlinear Feature Selective Networks (NFSN) that be able to select discriminative feature subsets while preserving their nonlinear structure by embedding a ℓ 2 , p -norm regularized hidden layer into designed continuous values regression networks. In addition, we propose an efficient optimization algorithm that joins back propagation algorithm and re-weighted optimization strategy to acquire derivative of all weights accurately. Experimental results on the nonlinear analog pulse signal and real-world datasets demonstrate the superiority of proposed method compared to some related methods on feature selection. Our source code available on: https://github.com/StevenWangNPU/NFSN. [ABSTRACT FROM AUTHOR]

• Performing object recognition with a new prototype of data glove. • Creation of a unique dataset containing 14 grasped objects. • Experiments (basic daily activities) conducted by 19 participants. • Comparing classical machine learning algorithms and one deep learning model. • Results of the object recognition is superior to 95%. In this paper, special attention is given to the hand. The literature provides solutions allowing the hand gestures recognition and/or object recognition for virtual reality, robotic applications, and so on. These solutions rely mainly on computer vision and data gloves. From this finding, we decided to develop our data glove prototype. The data glove is exploited to recognize common objects in the kitchen that the person can hold (e.g., hold a fork) while he/she performs basic daily activities such as drink a glass of water. The proposed approach is straightforward, cheap (∼ 260 $ in USD) and efficient (∼ 100%). Moreover, the designed data glove gives easy and direct access to the raw data provided by sensors. Besides, a comparison between classical machine learning algorithms (e.g., CART, Random Forest) and a deep neural network is given. Finally, the proposed prototype is described in a way that researchers can reproduce it for any applications involving the object recognition with the hand. [ABSTRACT FROM AUTHOR]

• We first exploit a deep framework with pattern theory to learn continuous temporal embedding. • We design a configuration proposal module to remove background information in video. • We explicitly formulate an incremental Viterbi algorithm for pattern theory. • The experimental results show that our CTE-PT overforms state-of-the-art methods. Visual Question Answering (VQA) is a challenging task in artificial intelligence and has received increasing attention from both the computer vision and the natural language processing communities. Joint embedding learning for it suffers from the background noise of images and text. Learning continuous temporal embedding is a potential solution to extract both visual elements and textual elements with a proper temporal structure. In this paper, we address a continuous temporal embedding model based on pattern theory (CTE-PT) and fully express the atomism and combinatory based on Grenander's pattern theory. First, we generate atomic actions from videos and note them as generators, which reflects the atomism in pattern theory. Second, we design a CTE-PT model to discover the discriminative configuration of videos, which reflects the combinatory of atomic actions in pattern theory. In the CTE-PT model, we design the configuration proposal module to remove some background information initially, and the configuration interpretation module to minimize the interpretive energy of continuous temporal embedding. We estimate the energy by each pair in this embedding sequence and optimize it based on the Viterbi algorithm. The experimental results show that our CTE-PT model outperforms the baseline C3D+LSTM model on Olympic Sports, UCF 101, and HMDB51 datasets, which proves the effectiveness of mining the common continuous temporal embedding as the class-specific configuration for activity discriminator. [Display omitted] [ABSTRACT FROM AUTHOR]

• Graph based video shot boundary detection using mutual information. • Graph based centrality measure to select a representative frame within a shot. • A center-surround model from spectral dissimilarity of two graphs. • MST based clustering in video similarity graph with new edge inadmissibility measure. [Display omitted] First-person video summarization has emerged as an important research problem for computer vision and multimedia communities. In this paper, we show how different graph representations can be developed for accurately summarizing first-person (egocentric) videos in a computationally efficient manner. Each frame in a video is first represented as a weighted graph. A shot boundary detection method using graph based mutual information is developed. We next construct a weighted graph for each shot. A representative frame from each shot is selected using a graph centrality measure. A new way of characterizing egocentric video frames using a graph based center-surround model is shown next. Here, each representative frame is modeled as a union of a center region (graph) and a surround region (graph). By exploiting spectral measures of dissimilarity between the two (center and surround) graphs, optimal center and surround regions are determined. Optimal regions for all frames within a shot are kept the same as that of the representative frame. Center-surround differences in entropy and optical flow values along with PHOG (Pyramidal HOG) features are extracted from each frame. All frames in a video are finally represented by another weighted graph, termed as a Video Similarity Graph (VSG). The frames are clustered by applying a Minimum Spanning Tree (MST) based approach with a new measure for inadmissible edges. Frames closest to the centroid of each cluster are captured to build the summary. Experimental evaluation on two benchmark datasets indicate the advantage of the proposed formulation. [ABSTRACT FROM AUTHOR]

• We examine deep learning model architectures applied to one-shot sensor-based HAR. • We introduce a modality-wise relational network to discover activity similarities. • We extensively investigate multi-head self-attention performance on one-shot HAR. • We construct explainable attention maps to display the modality-wise similarities. Deep learning concepts have been successfully transferred from the computer vision task to that of wearable human activity recognition (HAR) over the last few years. However, deep learning models require a large volume of annotated samples to be efficiently trained, while adding new activities results in training the whole network from scratch. In this paper, we study the use of one-shot learning techniques based on high-level features extracted by deep neural networks that rely on convolutional layers. Using these feature vectors as input we measure the similarity of two activities by computing their Euclidean distance, cosine similarity or applying self-attention to perceive the relations between the signals. We evaluate four different one-shot learning approaches using two publicly available HAR datasets, by keeping out of the training set several activity classes. Our results demonstrate that the model relying on modality-wise relational reasoning surpasses the other three, achieving 94.8% and 84.41% one-shot accuracy on UCL and PAMAP2 dataset respectively, while we demonstrate the model's sensitivity on fusing sensor modalities and provide explainable attention maps to display the modality-wise similarities. [ABSTRACT FROM AUTHOR]

• A fully residual convolutional neural network (FR-CNN) model is proposed for background subtraction. • The fully residual connection in FR-CNN helps to fuse the fine scale and coarse scale feature information efficiently. • A novel background modeling system is proposed. • The results show that the FR-CNN model outperforms the state-of-the-art methods. Background subtraction is an important step involved in solving computer vision problems. This paper proposes a novel background subtraction method with fully residual convolutional neural network (FR-CNN). This fully residual connection helps to fuse the fine scale and coarse scale feature information efficiently. The extracted non-handcrafted features are robust and promisingly efficient compared to the handcrafted features. Furthermore, the method uses temporal and spatial information for the background subtraction process. The optical flow image is used for extracting the temporal information. Additionally, a new background modeling technique is also proposed for the efficient background subtraction. The model is trained using the randomly selected 50 frames from each video sequence of the CDnet-2014 dataset and the FR-CNN model is evaluated by CDnet-2014 dataset. The results shown from the qualitative and quantitative analyses reveal that the FR-CNN model outperforms the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

• A novel graph-based non-maximal suppression scheme is introduced for detecting products on the rack. • A potential confidence score is defined for implementing graph-based non-maximal suppression approach. • A directed acyclic graph is strategically constructed for designing graph-based non-maximal suppression technique. • It is shown that graph-based non-maximal suppression method outperforms greedy non-maximal suppression. Identification of stacked retail products from the images of racks of supermarket is a challenging computer vision problem. Regions with convolutional neural network features (R-CNN) generate (mostly overlapped) region proposals around the products on the rack. Subsequently, the region proposals are classified using a convolutional neural network. In the end, R-CNN implements a greedy non-maximal suppression (greedy-NMS) for disambiguating the overlapping proposals. Greedy-NMS discards the proposals (with lower classification scores) that are overlapped with the proposal with higher classification score. This greedy approach often eliminates the (geometrically) better fitted region proposals with (marginally) lower classification scores. This paper introduces a novel graph-based non-maximal suppression (G-NMS) that removes this critical bottleneck of greedy-NMS by looking not only at the classification scores but also at the product classes of the overlapping region proposals. G-NMS first determines the potential confidence scores (pc-scores) of the region proposals by defining the groups of overlapping regions. Subsequently, a directed acyclic graph (DAG) is strategically constructed with the proposals utilizing their pc-scores and overlapping groups. Eventually the maximum weighted path of the DAG provides the products that are present in the rack. The results of our extensive experiments confirm that the proposed scheme is better up to around 7% on one large In-house and three benchmark datasets of retail products. Additionally, the efficacy of our proposed GNMS is also analyzed on four benchmark datasets for detecting generic objects. [ABSTRACT FROM AUTHOR]

In the recent years, the computer vision community has shown great interest on depth-based applications thanks to the performance and flexibility of the new generation of RGB-D imagery. In this paper, we present an efficient background subtraction algorithm based on the fusion of multiple region-based classifiers that processes depth and color data provided by RGB-D cameras. Foreground objects are detected by combining a region-based foreground prediction (based on depth data) with different background models (based on a Mixture of Gaussian algorithm) providing color and depth descriptions of the scene at pixel and region level. The information given by these modules is fused in a mixture of experts fashion to improve the foreground detection accuracy. The main contributions of the paper are the region-based models of both background and foreground, built from the depth and color data. The obtained results using different database sequences demonstrate that the proposed approach leads to a higher detection accuracy with respect to existing state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

• We propose a two-stage network, which includes Shape Initialization Network and cascaded Shape Refinement Networks. • A novel convolutional neural network, SIN is designed to initialize robust facial landmarks. • We present a multimodal deep learning structure to refine facial landmarks. • The superiority of the proposed method is demonstrated on 300-W dataset. Recently, face alignment plays a significant role in lots of computer vision tasks. Due to the various head poses, diverse expressions and partial occlusions, face alignment has still been a great challenge. In this paper, we propose a new Deep Shape Constrained Network (DSCN) for robust face alignment. Specifically, our DSCN consists of a Shape Initialization Network (SIN) and a series of Shape Refinement Networks (SRN) in a cascaded way. SIN takes the holistic facial image as the input and generates promising preliminary facial landmarks for SRN. Each SRN consists of two task-specific streams: shape stream aiming to learn the constraint of global face shape, and the texture stream used to extract robust face feature. Finally, the network automatically learns to combine these two streams in an early fusion approach. Experimental results show that the proposed method outperforms state-of-the-art approaches on 300-W dataset, which consists of LFPW, HELEN and AFW. [ABSTRACT FROM AUTHOR]

• Semantic segmentation is a challenging task in unstructured road environment. • Modification in original DeepLabV3+ is proposed for dense traffic scenario. • We propose to use dilated Xception as encoder for feature extraction. • A lightweight segmentation framework is also presented based on MobileNetV2. • Results are evaluated on India Driving Dataset having 26 classes. Semantic scene segmentation has become a key application in computer vision and is an essential part of intelligent transportation systems for complete scene understanding of the surrounding environment. While several methods based on deep fully Convolutional Neural Network (CNN) have been emerging, there are two main challenges: (i) They mainly focus on improvement of the accuracy than efficiency. (ii) They assume structured driving environment like in USA and Europe. While most of the current works focus on the well structured driving environment, we focus our research on India Driving Dataset (IDD) which contains data from unstructured traffic scenario. In this paper, we propose modifications in the DeepLabV3+ framework by using lower atrous rates in Atrous Spatial Pyramid Pooling (ASPP) module for dense traffic prediction. We propose to use dilated Xception network as the backbone for feature extraction. A lightweight segmentation framework is also presented by exploring the effectiveness of MobileNetV2 architecture, which achieves competitively high accuracy and is much smaller than other state-of-art architectures. The performance is evaluated in terms of mean Intersection over Union (mIoU) on 26 fine grained classes of IDD. Our proposed model with 24 M parameters achieves 68.41 mIoU on test set and efficient mobile model achieves mIoU of 61.6 by reducing the parameters to 2.2 M only. [ABSTRACT FROM AUTHOR]

• Survey on Positive Technology (PT) for elderly well-being. • Computer Vision and Machine Learning for PT. • Intelligent systems for capturing/modeling emotional, social, and cognitive human behaviour. • Relation between PT and Ambient-Assisted Living, GeronTechnology, and Affective Computing. In the last decades, given the necessity of assisting fragile citizens, of which elderly represent a significant portion, a considerable research effort has been devoted to the use of information and communication technologies (ICT) in daily living to promote activity, social connections, and independence. With similar purposes, in recent years psychologists proposed the novel paradigm of Positive Psychology (PP), the scientific study of positive human functioning and flourishing on multiple levels. The joint effort between ICT and PP has led to the definition of the emerging field of Positive Technology (PT), with the aim of developing technology consciously designed to foster well-being in individuals and groups. In this paper we review PT focusing on frameworks involving computer vision and machine learning for promoting cognitive, physical, emotional and social elderly well-being. Our discussion highlights a significant gap between theoretical needs and technological systems availability, suggesting future lines of research. [ABSTRACT FROM AUTHOR]

• We propose a novel block-diagonal covariance descriptor for image set classification. • The proposed descriptor significantly improves the performance as well as efficiency. • The proposed descriptor is effective with both shallow features and deep features. Image set classification has recently received much attention due to its various applications in pattern recognition and computer vision. To compare and match image sets, the major challenges are to devise an effective and efficient representation and to define a measure of similarity between image sets. In this paper, we propose a method for representing image sets based on block-diagonal Covariance Descriptors (CovDs). In particular, the proposed image set representation is in the form of non-singular covariance matrices, also known as Symmetric Positive Definite (SPD) matrices, that lie on Riemannian manifold. By dividing each image of an image set into square blocks of the same size, we compute the corresponding block CovDs instead of the global one. Taking the relative discriminative power of these block CovDs into account, a block-diagonal SPD matrix can be constructed to achieve a better discriminative capability. We extend the proposed approach to work with bidirectional CovDs and achieve a further boost in performance. The resulting block-diagonal SPD matrices combined with Riemannian metrics are shown to provide a powerful basis for image set classification. We perform an extensive evaluation on four datasets for several image set classification tasks. The experimental results demonstrate the effectiveness and efficiency of the proposed method. [ABSTRACT FROM AUTHOR]

• Facial landmark detection. • Cascaded regression robust to occlusions. • Accurate coordinate estimation from heatmaps. Convolutional Neural Nets (CNNs) have become the reference technology for many computer vision problems. Although CNNs for facial landmark detection are very robust, they still lack accuracy when processing images acquired in unrestricted conditions. In this paper we investigate the use of a cascade of Neural Net regressors to increase the accuracy of the estimated facial landmarks. To this end we append two encoder-decoder CNNs with the same architecture. The first net produces a set of heatmaps with a rough estimation of landmark locations. The second, trained with synthetically generated occlusions, refines the location of ambiguous and occluded landmarks. Finally, a densely connected layer with shared weights among all heatmaps, accurately regresses the landmark coordinates. The proposed approach achieves state-of-the-art results in 300W, COFW and WFLW that are widely considered the most challenging public data sets. [ABSTRACT FROM AUTHOR]

• We address the correspondence based point cloud registration problem with extremely high outlier ratio. • Potential inliers are chosen by checking the distance of corresponding point pairs. • Translation and rotation are separately estimated by utilizing transformation decomposition. • Deterministic and global optimal estimation is achieved by branch and bound algorithms. • Experiments show that the proposed method is more efficient than state of the art methods. Correspondence-based point cloud registration is widely used in the computer vision field and it is often solved by RANSAC, which is stable and fast when the outlier ratio is moderate or low. However, RANSAC will be very slow when the outlier ratio is extremely high, such as 99% or even higher, which is not rare especially when there are many repeated structures or the structure diversity is low in the scene. Outlier removal method was developed as a preprocessing step to improve the efficiency in very high outlier ratio, but its efficiency goes low when the outlier ratio goes down, such as lower than 95%. In this paper, we address this problem from another perspective. Instead of removing outliers, we select potential inliers by checking the consistency of distance of a pair of correspondences and estimate translation and rotation separately by transformation decomposition using the selected correspondence pairs. Experiments on both synthetic and real data show that the proposed method is fast in both high and low outlier ratios and it is significantly more efficient than RANSAC and outlier removal approach when the outlier ratio is extremely high. [ABSTRACT FROM AUTHOR]

• An attentive multi-stage CNN for the task of crowd counting. • A hierarchical density estimator for mining semantic features. • A soft attentive mechanism for focusing on foreground of interest. Crowd counting is an important problem in computer vision, whose application can be found in a wide range of tasks. Although this problem has been well studied, how to effectively deal with scale variations and perspective distortions is still a challenge. High-quality crowd density map depends heavily on how well these problems are solved. In this paper, we propose a novel network architecture called Attentive Multi-stage CNN for Crowd Counting (AMCNN). The AMCNN contains two subnetworks, i.e., hierarchical density estimator(HDE) and auxiliary count classifier (AUCC). The HDE adopts a hierarchical strategy to mine semantic features in a coarse-to-fine manner to tackle the problem of scale changes and perspective distortions. And the obtained composite features are used to generate the final density map. In addition, to further improve the density map quality, a soft attention mechanism is integrated into the AMCNN to distinct the foreground and the background. Furthermore, the AUCC is employed to achieve the count classification task, which is complementary to the task of density estimation. We evaluate our model on three public datasets: ShanghaiTech, UCF_CC_50 and Mall. Extensive experiments demonstrate that our counting model is on par with some state-of-the-art methods. Source code will be released at: https://github.com/wxq-ahu/crowd-count-amcnn. [ABSTRACT FROM AUTHOR]

• Feature extraction alone does not increase the accuracy of defect detection. • The usage of Neural Networks increases the accuracy with the feature extraction method. • Nine different training algorithms tested and Levenberg–Marquardt is selected based on its high accuracy. • Use of Optimized Neural Networks further enhanced the accuracy in the detection of almost all types of defects. • Developed Optimized Neural Network classifier using GA, PSO, ABC. ABC-ANN outperforms the other two. It is essential to identify the defects on leather sheets in leather industries as part of quality control. Manual inspection is often considered inconsistent due to a lack of accuracy and time constraints. Automation of the process using a machine vision system gives high accuracy and consistency in the classification of the leather sheets basing on their quality. In this paper, a technique to detect defects on wet-blue leather and classifying them using artificial neural networks (ANN) is proposed. Features of several defects on the leather are extracted using a GreyLevelCo-occurrence matrix(GLCM) and GreyLevelRunLengthMatrix(GLRLM). The obtained features are given to a multi-layer perceptron with a Levenberg–Marquardt (LM) algorithm. The numeral of HiddenLayers (N HL) and the numeral of neurons in each HiddenLayer(NN HL) of the network are optimized using a genetic algorithm(GA), particle swarm optimization (PSO) and artificial bee colony optimization(ABC). The simulation-based experimentation is carried out in MATLAB. Classifier outputs are analyzed using performance metrics like specificity, accuracy, and sensitivity. The LM-ANN based model with ABC optimization outperforms the other two techniques in classifying the defects of wet blue leather with performance metrics with 98.73% of Specificity, 97.85% of Accuracy and 94.14% of Sensitivity. [ABSTRACT FROM AUTHOR]

• The covariant local feature detector is improved by complementary information. • Keypoints are detected by incorporating the confidence into the predicted positions. • The proposed method is a general framework fusing two different keypoint detectors. • State of the art performance has been obtained on four benchmarks. Local feature detection is a fundamental problem in computer vision. Recently, the research of local feature detection has been switched from handcrafted methods to learning based ones, especially deep learning based ones. A recent successful deep learning based feature detector is the covariant local feature detector that conducts keypoint detection by predicting the transformation of keypoints from nearby pixels. Although this method adopts a new detection framework compared to those methods by computing the keypoint's likelihood, it treats each pixel equally which may incorrectly detect unstable keypoints. On the other hand, other methods computing the keypoint probability could capture different evidence for keypoint detection as well as provide a natural weight for each prediction in the covariant detector. So, fusing information from other detectors into the covariant detector could improve its performance. Under this motivation, this paper proposes an improved covariant local feature detector by fusing feature information obtained from another detector, which is served as a confidence to guide the voting procedure when converting the predicted transformations into a meaningful score map for keypoint detection. In this way, the fused information can enhance the features that are considered to be good and weaken those unstable features. The proposed method is evaluated on four widely used benchmarks and consistent performance improvement over previous works is observed. [ABSTRACT FROM AUTHOR]

• An automatic region of interest segmentation method • A set of rules as a metric to measure the region of interest segmentation results • Analysis of thermal matrices as new data source for feature extraction • Design of experiments in a more robust manner taking into account most of the images in the database • Genetic algorithm for hyperparameter tunning for the Artificial Neural Network Breast thermography images are a new type of data that has been analyzed in recent years in order to detect abnormalities, which can lead to a future breast cancer. This paper proposes a methodology for breast thermal image classification, which is useful in Computer-Aided Detection Systems. The main contribution is an automatic method to segment the region of interest (ROI) based on local operations, local analysis, interpolation and statistical operators. For our experiments, we used an image database that is widely used in this research area, obtaining accuracy results between 90.17% and 98.33%, which are competitive with respect to related works. [ABSTRACT FROM AUTHOR]

• We propose a family of local-global nested graph kernels based on nested representations. • The nested aligned kernel is based on the dynamic time warping inspired measures. • The nested reproducing kernel is based on the basic H 1-reproducing kernel. • Both the nested kernels simultaneously reflect the local and global graph characteristics. • Experiments demonstrate the effectiveness and efficiency of the proposed graph kernels. In this paper, we propose two novel local-global nested graph kernels, namely the nested aligned kernel and the nested reproducing kernel, drawing on depth-based complexity traces. Both of the nested kernels gauge the nested depth complexity trace through a family of K -layer expansion subgraphs rooted at the centroid vertex, i.e., the vertex with minimum shortest path length variance to the remaining vertices. Specifically, for a pair of graphs, we commence by computing the centroid depth-based complexity traces rooted at the centroid vertices. The first nested kernel is defined by measuring the global alignment kernel, which is based on the dynamic time warping framework, between the complexity traces. Since the required global alignment kernel incorporates the whole spectrum of alignment costs between the complexity traces, this nested kernel can provide rich statistic measures. The second nested kernel, on the other hand, is defined by measuring the basic reproducing kernel between the complexity traces. Since the associated reproducing kernel only requires time complexity O (1), this nested kernel has very low computational complexity. We theoretically show that both of the proposed nested kernels can simultaneously reflect the local and global graph characteristics in terms of the nested complexity traces. Experiments on standard graph datasets abstracted from bioinformatics and computer vision databases demonstrate the effectiveness and efficiency of the proposed graph kernels. [ABSTRACT FROM AUTHOR]