Search

Along with current multi-scale based detectors, Feature Aggregation and Enhancement (FAE) modules have shown superior performance gains for cutting-edge object detection. However, these hand-crafted FAE modules show inconsistent improvements on face detection, which is mainly due to the significant distribution difference between its training and applying corpus, COCO vs. WIDER Face. To tackle this problem, we essentially analyse the effect of data distribution, and consequently propose to search an effective FAE architecture, termed AutoFAE by a differentiable architecture search, which outperforms all existing FAE modules in face detection with a considerable margin. Upon the found AutoFAE and existing backbones, a supernet is further built and trained, which automatically obtains a family of detectors under the different complexity constraints. Extensive experiments conducted on popular benchmarks, WIDER Face and FDDB, demonstrate the state-of-the-art performance-efficiency trade-off for the proposed automatic and scalable face detector (ASFD) family. In particular, our strong ASFD-D6 outperforms the best competitor with AP 96.7/96.2/92.1 on WIDER Face test, and the lightweight ASFD-D0 costs about 3.1 ms, more than 320 FPS, on the V100 GPU with VGA-resolution images., ACM MM2021

In this paper, we place the atomic action detection problem into a Long-Short Term Context (LSTC) to analyze how the temporal reliance among video signals affect the action detection results. To do this, we decompose the action recognition pipeline into short-term and long-term reliance, in terms of the hypothesis that the two kinds of context are conditionally independent given the objective action instance. Within our design, a local aggregation branch is utilized to gather dense and informative short-term cues, while a high order long-term inference branch is designed to reason the objective action class from high-order interaction between actor and other person or person pairs. Both branches independently predict the context-specific actions and the results are merged in the end. We demonstrate that both temporal grains are beneficial to atomic action recognition. On the mainstream benchmarks of atomic action detection, our design can bring significant performance gain from the existing state-of-the-art pipeline. The code of this project can be found at [this url](https://github.com/TencentYoutuResearch/ActionDetection-LSTC), ACM Multimedia 2021

Recently, most siamese network based trackers locate targets via object classification and bounding-box regression. Generally, they select the bounding-box with maximum classification confidence as the final prediction. This strategy may miss the right result due to the accuracy misalignment between classification and regression. In this paper, we propose a novel siamese tracking algorithm called SiamRCR, addressing this problem with a simple, light and effective solution. It builds reciprocal links between classification and regression branches, which can dynamically re-weight their losses for each positive sample. In addition, we add a localization branch to predict the localization accuracy, so that it can work as the replacement of the regression assistance link during inference. This branch makes the training and inference more consistent. Extensive experimental results demonstrate the effectiveness of SiamRCR and its superiority over the state-of-the-art competitors on GOT-10k, LaSOT, TrackingNet, OTB-2015, VOT-2018 and VOT-2019. Moreover, our SiamRCR runs at 65 FPS, far above the real-time requirement., The 30th International Joint Conference on Artificial Intelligence (IJCAI 2021)

In this work, we propose a high fidelity face swapping method, called HifiFace, which can well preserve the face shape of the source face and generate photo-realistic results. Unlike other existing face swapping works that only use face recognition model to keep the identity similarity, we propose 3D shape-aware identity to control the face shape with the geometric supervision from 3DMM and 3D face reconstruction method. Meanwhile, we introduce the Semantic Facial Fusion module to optimize the combination of encoder and decoder features and make adaptive blending, which makes the results more photo-realistic. Extensive experiments on faces in the wild demonstrate that our method can preserve better identity, especially on the face shape, and can generate more photo-realistic results than previous state-of-the-art methods. Code is available at: https://johann.wang/HifiFace

Recent advances in image inpainting have shown impressive results for generating plausible visual details on rather simple backgrounds. However, for complex scenes, it is still challenging to restore reasonable contents as the contextual information within the missing regions tends to be ambiguous. To tackle this problem, we introduce pretext tasks that are semantically meaningful to estimating the missing contents. In particular, we perform knowledge distillation on pretext models and adapt the features to image inpainting. The learned semantic priors ought to be partially invariant between the high-level pretext task and low-level image inpainting, which not only help to understand the global context but also provide structural guidance for the restoration of local textures. Based on the semantic priors, we further propose a context-aware image inpainting model, which adaptively integrates global semantics and local features in a unified image generator. The semantic learner and the image generator are trained in an end-to-end manner. We name the model SPL to highlight its ability to learn and leverage semantic priors. It achieves the state of the art on Places2, CelebA, and Paris StreetView datasets., Accepted by IJCAI 2021

Temporal action localization is an important yet challenging task in video understanding. Typically, such a task aims at inferring both the action category and localization of the start and end frame for each action instance in a long, untrimmed video.While most current models achieve good results by using pre-defined anchors and numerous actionness, such methods could be bothered with both large number of outputs and heavy tuning of locations and sizes corresponding to different anchors. Instead, anchor-free methods is lighter, getting rid of redundant hyper-parameters, but gains few attention. In this paper, we propose the first purely anchor-free temporal localization method, which is both efficient and effective. Our model includes (i) an end-to-end trainable basic predictor, (ii) a saliency-based refinement module to gather more valuable boundary features for each proposal with a novel boundary pooling, and (iii) several consistency constraints to make sure our model can find the accurate boundary given arbitrary proposals. Extensive experiments show that our method beats all anchor-based and actionness-guided methods with a remarkable margin on THUMOS14, achieving state-of-the-art results, and comparable ones on ActivityNet v1.3. Code is available at https://github.com/TencentYoutuResearch/ActionDetection-AFSD., Accepted by CVPR2021

Super-Resolution (SR) is a fundamental computer vision task that aims to obtain a high-resolution clean image from the given low-resolution counterpart. This paper reviews the NTIRE 2021 Challenge on Video Super-Resolution. We present evaluation results from two competition tracks as well as the proposed solutions. Track 1 aims to develop conventional video SR methods focusing on the restoration quality. Track 2 assumes a more challenging environment with lower frame rates, casting spatio-temporal SR problem. In each competition, 247 and 223 participants have registered, respectively. During the final testing phase, 14 teams competed in each track to achieve state-of-the-art performance on video SR tasks.

Most of the existing deep learning-based dehazing methods are trained and evaluated on the image dehazing datasets, where the dehazed images are generated by only exploiting the information from the corresponding hazy ones. On the other hand, video dehazing algorithms, which can acquire more satisfying dehazing results by exploiting the temporal redundancy from neighborhood hazy frames, receive less attention due to the absence of the video dehazing datasets. Therefore, we propose the first REal-world VIdeo DEhazing (REVIDE) dataset which can be used for the supervised learning of the video dehazing algorithms. By utilizing a well-designed video acquisition system, we can capture paired real-world hazy and haze-free videos that are perfectly aligned by recording the same scene (with or without haze) twice. Considering the challenge of exploiting temporal redundancy among the hazy frames, we also develop a Confidence Guided and Improved Deformable Network (CG-IDN) for video dehazing. The experiments demonstrate that the hazy scenes in the REVIDE dataset are more realistic than the synthetic datasets and the proposed algorithm also performs favorably against state-of-the-art dehazing methods.

Non-parametric face modeling aims to reconstruct 3D face only from images without shape assumptions. While plausible facial details are predicted, the models tend to over-depend on local color appearance and suffer from ambiguous noise. To address such problem, this paper presents a novel Learning to Aggregate and Personalize (LAP) framework for unsupervised robust 3D face modeling. Instead of using controlled environment, the proposed method implicitly disentangles ID-consistent and scene-specific face from unconstrained photo set. Specifically, to learn ID-consistent face, LAP adaptively aggregates intrinsic face factors of an identity based on a novel curriculum learning approach with relaxed consistency loss. To adapt the face for a personalized scene, we propose a novel attribute-refining network to modify ID-consistent face with target attribute and details. Based on the proposed method, we make unsupervised 3D face modeling benefit from meaningful image facial structure and possibly higher resolutions. Extensive experiments on benchmarks show LAP recovers superior or competitive face shape and texture, compared with state-of-the-art (SOTA) methods with or without prior and supervision., Comment: CVPR 2021 Oral, 11 pages, 9 figures

The conversion of grassland to cropland is one of the major changes in land use, and it accelerates both soil erosion and the loss of soil organic carbon (SOC). However, the general patterns of SOC loss after grassland cultivation are rarely assessed, and the potential mechanisms remain unclear. Here, a meta-analysis of 81 case studies was performed to show that SOC decreased with soil depths of 0–60 cm after grassland conversion, but no significant differences were found at depths >60 cm. SOC also declined significantly with the duration of grassland conversion. The response ratio of SOC changes tended to reach equilibrium after 20 years of grassland cropping. Our results indicate that reduction in SOC mainly depended on changes in precipitation, soil physical-chemical properties and soil microbes. These conclusions highlight the importance of improving the accuracy of predictions on SOC losses and on the global carbon cycle in the face of land-use changes worldwide.

Few-shot learning (FSL), which aims to recognise new classes by adapting the learned knowledge with extremely limited few-shot (support) examples, remains an important open problem in computer vision. Most of the existing methods for feature alignment in few-shot learning only consider image-level or spatial-level alignment while omitting the channel disparity. Our insight is that these methods would lead to poor adaptation with redundant matching, and leveraging channel-wise adjustment is the key to well adapting the learned knowledge to new classes. Therefore, in this paper, we propose to learn a dynamic alignment, which can effectively highlight both query regions and channels according to different local support information. Specifically, this is achieved by first dynamically sampling the neighbourhood of the feature position conditioned on the input few shot, based on which we further predict a both position-dependent and channel-dependent Dynamic Meta-filter. The filter is used to align the query feature with position-specific and channel-specific knowledge. Moreover, we adopt Neural Ordinary Differential Equation (ODE) to enable a more accurate control of the alignment. In such a sense our model is able to better capture fine-grained semantic context of the few-shot example and thus facilitates dynamical knowledge adaptation for few-shot learning. The resulting framework establishes the new state-of-the-arts on major few-shot visual recognition benchmarks, including miniImageNet and tieredImageNet., accepted by CVPR2021

With rapid urbanization in China, the dramatic land-use changes are one of the most prominent features that have substantially affected the land ecosystems, thus seriously threatening sustainable development. However, current studies have focused more on evaluating the economic efficiency of land-use, while the loss and degradation of ecosystem services are barely considered. To address these issues, this study first proposed a land use-based input&ndash, output index system, incorporating the impact on ecosystem services value (ESV), and then by taking 30 provinces in China as a case study. We further employed the super-efficiency slacks-based model (Super-SBM) and the Stochastic Impacts by Regression on Population, Affluence and Technology (STIRPAT) model to explore the spatial&ndash, temporal changes and driving factors of the evaluated land-use eco-efficiency. We found that the evaluated ESV was 28.09 trillion yuan (at the price of 2000) in 2015, and that the total ESV experienced an inverted U-shaped trend during 2000&ndash, 2015.The average land-use eco-efficiency exhibited a downward trend from 0.87 in 2000 to 0.68 in 2015 with distinct regional differences by taking into account the ESV. Our results revealed that northeastern region had the highest efficiency, followed by the eastern, western, and central region of China. Finally, we identified a U-shaped relationship between the eco-efficiency and land urbanization, and found that technological innovation made great contributions to the improvement of the eco-efficiency. These findings highlight the importance of the ESV in the evaluation of land-use eco-efficiency. Future land development and management should pay additional attention to the land ecosystems, especially the continuous supply of human well-being related ecosystem services.

Human motion prediction aims to predict future 3D skeletal sequences by giving a limited human motion as inputs. Two popular methods, recurrent neural networks and feed-forward deep networks, are able to predict rough motion trend, but motion details such as limb movement may be lost. To predict more accurate future human motion, we propose an Adversarial Refinement Network (ARNet) following a simple yet effective coarse-to-fine mechanism with novel adversarial error augmentation. Specifically, we take both the historical motion sequences and coarse prediction as input of our cascaded refinement network to predict refined human motion and strengthen the refinement network with adversarial error augmentation. During training, we deliberately introduce the error distribution by learning through the adversarial mechanism among different subjects. In testing, our cascaded refinement network alleviates the prediction error from the coarse predictor resulting in a finer prediction robustly. This adversarial error augmentation provides rich error cases as input to our refinement network, leading to better generalization performance on the testing dataset. We conduct extensive experiments on three standard benchmark datasets and show that our proposed ARNet outperforms other state-of-the-art methods, especially on challenging aperiodic actions in both short-term and long-term predictions.

Multiple Object Tracking (MOT) is an important task in computer vision. MOT is still challenging due to the occlusion problem, especially in dense scenes. Following the tracking-by-detection framework, we propose the Box-Plane Matching (BPM) method to improve the MOT performacne in dense scenes. First, we design the Layer-wise Aggregation Discriminative Model (LADM) to filter the noisy detections. Then, to associate remaining detections correctly, we introduce the Global Attention Feature Model (GAFM) to extract appearance feature and use it to calculate the appearance similarity between history tracklets and current detections. Finally, we propose the Box-Plane Matching strategy to achieve data association according to the motion similarity and appearance similarity between tracklets and detections. With the effectiveness of the three modules, our team achieves the 1st place on the Track-1 leaderboard in the ACM MM Grand Challenge HiEve 2020., Comment: ACM Multimedia 2020 GC paper. ACM Multimedia Grand Challenge HiEve 2020 Track-1 Winner

Unsupervised learning of optical flow, which leverages the supervision from view synthesis, has emerged as a promising alternative to supervised methods. However, the objective of unsupervised learning is likely to be unreliable in challenging scenes. In this work, we present a framework to use more reliable supervision from transformations. It simply twists the general unsupervised learning pipeline by running another forward pass with transformed data from augmentation, along with using transformed predictions of original data as the self-supervision signal. Besides, we further introduce a lightweight network with multiple frames by a highly-shared flow decoder. Our method consistently gets a leap of performance on several benchmarks with the best accuracy among deep unsupervised methods. Also, our method achieves competitive results to recent fully supervised methods while with much fewer parameters., Accepted to CVPR 2020, https://github.com/lliuz/ARFlow

Recent state-of-the-art super-resolution methods have achieved impressive performance on ideal datasets regardless of blur and noise. However, these methods always fail in real-world image super-resolution, since most of them adopt simple bicubic downsampling from high-quality images to construct Low-Resolution (LR) and High-Resolution (HR) pairs for training which may lose track of frequency-related details. To address this issue, we focus on designing a novel degradation framework for real- world images by estimating various blur kernels as well as real noise distributions. Based on our novel degradation framework, we can acquire LR images sharing a common domain with real-world images. Then, we propose a real- world super-resolution model aiming at better perception. Extensive experiments on synthetic noise data and real- world images demonstrate that our method outperforms the state-of-the-art methods, resulting in lower noise and better visual quality. In addition, our method is the winner of NTIRE 2020 Challenge on both tracks of Real-World Super-Resolution, which significantly outperforms other competitors by large margins.

Existing Multiple-Object Tracking (MOT) methods either follow the tracking-by-detection paradigm to conduct object detection, feature extraction and data association separately, or have two of the three subtasks integrated to form a partially end-to-end solution. Going beyond these sub-optimal frameworks, we propose a simple online model named Chained-Tracker (CTracker), which naturally integrates all the three subtasks into an end-to-end solution (the first as far as we know). It chains paired bounding boxes regression results estimated from overlapping nodes, of which each node covers two adjacent frames. The paired regression is made attentive by object-attention (brought by a detection module) and identity-attention (ensured by an ID verification module). The two major novelties: chained structure and paired attentive regression, make CTracker simple, fast and effective, setting new MOTA records on MOT16 and MOT17 challenge datasets (67.6 and 66.6, respectively), without relying on any extra training data. The source code of CTracker can be found at: github.com/pjl1995/CTracker.

Motivated by the previous success of Two-Dimensional Convolutional Neural Network (2D CNN) on image recognition, researchers endeavor to leverage it to characterize videos. However, one limitation of applying 2D CNN to analyze videos is that different frames of a video share the same 2D CNN kernels, which may result in repeated and redundant information utilization, especially in the spatial semantics extraction process, hence neglecting the critical variations among frames. In this paper, we attempt to tackle this issue through two ways. 1) Design a sequential channel filtering mechanism, i.e., Progressive Enhancement Module (PEM), to excite the discriminative channels of features from different frames step by step, and thus avoid repeated information extraction. 2) Create a Temporal Diversity Loss (TD Loss) to force the kernels to concentrate on and capture the variations among frames rather than the image regions with similar appearance. Our method is evaluated on benchmark temporal reasoning datasets Something-Something V1 and V2, and it achieves visible improvements over the best competitor by \(2.4\%\) and \(1.3\%\), respectively. Besides, performance improvements over the 2D-CNN-based state-of-the-arts on the large-scale dataset Kinetics are also witnessed.

Existing facial attribute editing methods typically employ an encoder-decoder architecture where the attribute information is expressed as a conditional one-hot vector spatially concatenated with the image or intermediate feature maps. However, such operations only learn the local semantic mapping but ignore global facial statistics. In this work, we focus on solving this issue by editing the channel-wise global information denoted as the style feature. We develop a style skip connection based generative adversarial network, referred to as SSCGAN which enables accurate facial attribute manipulation. Specifically, we inject the target attribute information into multiple style skip connection paths between the encoder and decoder. Each connection extracts the style feature of the latent feature maps in the encoder and then performs a residual learning based mapping function in the global information space guided by the target attributes. In the following, the adjusted style feature will be utilized as the conditional information for instance normalization to transform the corresponding latent feature maps in the decoder. In addition, to avoid the vanishing of spatial details (e.g. hairstyle or pupil locations), we further introduce the skip connection based spatial information transfer module. Through the global-wise style and local-wise spatial information manipulation, the proposed method can produce better results in terms of attribute generation accuracy and image quality. Experimental results demonstrate the proposed algorithm performs favorably against the state-of-the-art methods.

Human pose estimation is the task of localizing body keypoints from still images. The state-of-the-art methods suffer from insufficient examples of challenging cases such as symmetric appearance, heavy occlusion and nearby person. To enlarge the amounts of challenging cases, previous methods augmented images by cropping and pasting image patches with weak semantics, which leads to unrealistic appearance and limited diversity. We instead propose Semantic Data Augmentation (SDA), a method that augments images by pasting segmented body parts with various semantic granularity. Furthermore, we propose Adversarial Semantic Data Augmentation (ASDA), which exploits a generative network to dynamically predict tailored pasting configuration. Given off-the-shelf pose estimation network as discriminator, the generator seeks the most confusing transformation to increase the loss of the discriminator while the discriminator takes the generated sample as input and learns from it. The whole pipeline is optimized in an adversarial manner. State-of-the-art results are achieved on challenging benchmarks. The code has been publicly available at https://github.com/Binyr/ASDA.

In this paper, we propose a novel face detection network with three novel contributions that address three key aspects of face detection, including better feature learning, progressive loss design and anchor assign based data augmentation, respectively. First, we propose a Feature Enhance Module (FEM) for enhancing the original feature maps to extend the single shot detector to dual shot detector. Second, we adopt Progressive Anchor Loss (PAL) computed by two different sets of anchors to effectively facilitate the features. Third, we use an Improved Anchor Matching (IAM) by integrating novel anchor assign strategy into data augmentation to provide better initialization for the regressor. Since these techniques are all related to the two-stream design, we name the proposed network as Dual Shot Face Detector (DSFD). Extensive experiments on popular benchmarks, WIDER FACE and FDDB, demonstrate the superiority of DSFD over the state-of-the-art face detectors., Comment: Camera-ready version of DSFD for CVPR 2019. Code is available at: https://github.com/TencentYoutuResearch/FaceDetection-DSFD

The sequential hard and soft failure mode, a typical failure phenomenon, involves degradation that starts after an initiation period. According to this situation, a random initiation effect is introduced to the normal degradation stage. Intuitively, a lifetime delayed degradation process (LDDP) provides a general framework for this typical complex failure mode. In the present study, general reliability inference approaches involving the joint likelihood function are developed for the LDDP with repeated measurements, according to the expectation maximization (EM) and stochastic EM algorithms, along with numerical simulations and practical application based on real data. Additionally, statistical inferences were obtained using a bootstrap method on the basis of parameter estimations. The proposed method was compared with the traditional two-step strategy under different sample sizes and inspection frequencies and exhibited enhanced performance.

In this work, a miniaturized sensor was integrated on fiber surface and developed for oxygen determination through evanescent-wave induced fluorescence quenching. The sensor was designed by using light emitting diode (LED) as light source and optical fiber as light transmission element. Tris(2,2′-bipyridyl) ruthenium ([Ru(bpy)3]2+) fluorophore was immobilized in the organically modified silicates (ORMOSILs) film and coated onto the fiber surface. When light propagated by total internal reflection (TIR) in the fiber core, evanescent wave could be produced on the fiber surface and excite [Ru(bpy)3]2+ fluorophore to produce fluorescence emission. Then oxygen could be determinated by its quenching effect on the fluorescence and its concentration could be evaluated according to Stern–Volumer model. Through integrating evanescent wave excitation and fluorescence quenching on fiber surface, the sensor was successfully miniaturized and exhibit improved performances of high sensitivity (1.4), excellent repeatability (1.2%) and fast analysis (12 s) for oxygen determination. The sensor provided a newly portable method for in-situ and real-time measurement of oxygen and showed potential for practical oxygen analysis in different application fields. Furthermore, the fabrication of this sensor provides a miniaturized and portable detection platform for species monitoring by simple modular design.

Methane (CH4) emissions from ruminants should be accounted for the natural grazed rangeland ecosystems when devising greenhouse gas budget inventory, in particular, their contribution to global warming. In this study, CH4 emission from sheep respiration at different grazing intensities (light grazing, 0.75 sheep/ha, LG; moderate grazing, 1.50 sheep/ha, MG; and heavy grazing, 2.25 sheep/ha, HG) and in sheepfolds were evaluated in a desert grassland of Inner Mongolia. Results indicated that daily CH4 emission from sheep was not significantly different between treatments. When CH4 emission was expressed emission per 100g daily, there was a significant difference of LG vs HG and MG vs HG, with the values of 15.64g, 20.00g and 28.63g for LG, MG and HG, respectively, during the grazing season. There was no significant difference among CH4 fluxes in sheepfolds (mean 39.0 ug m-1 h-1). Considering CH4 emissions from the grazing ecosystem, net CH4 emissions from LG, MG and HG plots were -18.33, -1.91 and 21.19 g/ha/day, respectively. The digestibility of forage had a positive correlation with CH emission expressed on daily and metabolic body weight basis. It is concluded that MG will improve the balance between CH emission from grassland and grazing livestock in the desert grasslands of Inner Mongolia.

In recent years, heatmap regression based models have shown their effectiveness in face alignment and pose estimation. However, Conventional Heatmap Regression (CHR) is not accurate nor stable when dealing with high-resolution facial videos, since it finds the maximum activated location in heatmaps which are generated from rounding coordinates, and thus leads to quantization errors when scaling back to the original high-resolution space. In this paper, we propose a Fractional Heatmap Regression (FHR) for high-resolution video-based face alignment. The proposed FHR can accurately estimate the fractional part according to the 2D Gaussian function by sampling three points in heatmaps. To further stabilize the landmarks among continuous video frames while maintaining the precise at the same time, we propose a novel stabilization loss that contains two terms to address time delay and non-smooth issues, respectively. Experiments on 300W, 300-VW and Talking Face datasets clearly demonstrate that the proposed method is more accurate and stable than the state-of-the-art models., Comment: Accepted to AAAI 2019. 8 pages, 7 figures

The study of face alignment has been an area of intense research in computer vision, with its achievements widely used in computer graphics applications. The performance of various face alignment methods is often image-dependent or somewhat random because of their own strategy. This study aims to develop a method that can select an input image with good face alignment results from many results produced by a single method or multiple ones. The task is challenging because different face alignment results need to be evaluated without any ground truth. This study addresses this problem by designing a feasible feature extraction scheme to measure the quality of face alignment results. The feature is then used in various machine learning algorithms to rank different face alignment results. Our experiments show that our method is promising for ranking face alignment results and is able to pick good face alignment results, which can enhance the overall performance of a face alignment method with a random strategy. We demonstrate the usefulness of our ranking-enhanced face alignment algorithm in two practical applications: face cartoon stylization and digital face makeup.

In this paper, we consider the face verification problem, which is to determine whether two face images belong to the same subject or not. Although many research efforts have been focused on this problem, it still remains a challenging problem due to large intra-personal variations in imaging conditions, such as illumination, pose, expression, and occlusion. Our proposed method is based on the idea that we would like the similarity between positive pairs larger than negative pairs, and obtain a similarity estimation of two images. We construct our decision function by incorporating bilinear similarity and Mahalanobis distance to the sigmoid function. The constructed decision function makes our method discriminative for inter-personal differences and invariant to intra-personal variations such as pose/lighting/expression. What is more, our formulated objective function is convex, which guarantees global minimum. Our method belongs to nonlinear metric which is more robust to handle heterogeneous data than linear metric. We evaluate our proposed verification method on the challenging labeled faces in the wild (LFW) database. Experimental results demonstrate that the proposed method outperforms state-of-the-art methods such as Joint Bayesian under the unrestricted setting of LFW.

When tracking facial landmarks in a video, existing face alignment methods seem not to be so accurate as they are employed frame by frame. This paper shows that zigzags on the trace of estimated landmarks make the estimation error perceptible. The reason why the zigzags occur is that the increment of landmark position is comparable to the estimation error and the frames are processed individually. In this paper, we train a CNN facial landmark detection model as a baseline method, and develop a post-processing algorithm to address the zigzag problem. The CNN model achieves state-of-the-art performance on the 300-W dataset. The post-processing algorithm based on ridge regression exploits correlation among adjacent frames to transform random errors into bias errors. As a result zigzags are eliminated, and the traces of landmarks look smoother while the mean error remains unchanged or even slightly decreases. Our algorithm runs on a mobile device (iPhone 5s) at 150 Fps. Extensive experiments conducted on the 300-VW dataset demonstrate the effectiveness of the proposed algorithm.

Dynamic Spectrum Access technology is implemented to help making use of the limited spectrum sources in wireless communication system due to its unique characteristics. However, some threats to this special network were noted with the same reasons. The Primary User Emulation Attack, for example, is a classic issue in Dynamic Spectrum Access networks. In this paper, a continuous Markov Chain Model is used to depict the behavior of dynamic spectrum access with network attack, analyze the characteristics of this network and discuss the performance changes under malicious user's influence with different parameters. The results verify that the modeling and analysis are valid.

This paper proposes a multi-wing chaotic system based on an even-symmetric multi-segment quadratic function. The dynamic characteristics of the new chaotic system are analyzed in detail. The system is implemented by the circuit using Multisim. The results of Matlab simulation and Multisim simulation are consistent with each other, which verifies the existence of the multi-wing attractors in the new system. The system is applied to a secure communication implementation based on the active-passive drive-response synchronization method. The simulation results for secure communication application demonstrate that the multi-wing chaotic system can successfully encrypt and decrypt message signals.

DNA sequencing is quickly evolving into a mobile technology with a need for power efficient compute services. This paper describes such a service in the form of a VLSI base caller chip for nanopore-based 3rd generation DNA sequencers. The base caller achieves a throughput equivalent to 10 human genomes per hour with a power consumption of 200 mW. This represents a 100× power reduction over a general-purpose solution with a 1000× improvement in throughput.

In view of the current low level of electrification in global and Chinese transportation, this paper selects five key areas to improve the level of electrification, such as electric vehicle, urban rail transit, electrified railway, ship shore power and airport bridge equipment, carries out technical and economic analysis, and studies and evaluates the future development trend of the five technologies. Finally, according to the situation in 2020 and 2030 respectively, this paper carried out the calculation of electrification improvement potential. The research results provide a key path for improving the electrification level in the field of transportation, and analyze the feasibility and future potential of the path, providing a reference for the global and Chinese governments to improve the electrification level in the field of transportation.

The contribution of livestock excreta to greenhouse gases (GHGs) emissions by sheep grazing in a typical steppe system in Guyuan county, Hebei province of the People's Republic of China was evaluated. Changes of methane (CH4) and carbon dioxide (CO2) fluxes from urine and dung patches excreted by sheep on grassland were measured for the first 144 h during July, August, September and October in 2011. CH4 fluxes from dung patches significantly increased (P

DNA sequencing based on nanopore sensors is now entering the marketplace. The ability to interface this technology to established CMOS microelectronics promises significant improvements in functionality and miniaturization. Among the key functions to benefit from this interface will be basecalling, the conversion of raw electronic molecular signatures to nucleotide sequence predictions. This paper presents the design and performance potential of custom CMOS base-callers embedded alongside nanopore sensors. A basecalliing architecture implemented in 32-nm technology is discussed with the ability to process the equivalent of 20 human genomes per day in real-time at a power density of 5 W/cm2 assuming a 3-mer nanopore sensor.

Deep convolutional network has been widely used in face recognition while not often used in face alignment. One of the most important reasons of this is the lack of training images annotated with landmarks due to fussy and time-consuming annotation work. To overcome this problem, we propose a novel data augmentation strategy. And we design an innovative training algorithm with adaptive learning rate for two iterative procedures, which helps the network to search an optimal solution. Our convolutional network can learn global high-level features and directly predict the coordinates of facial landmarks. Extensive evaluations show that our approach outperforms state-of-the-art methods especially in the condition of complex occlusion, pose, illumination and expression variations.

We compare water tunnel experiment and 2D vortex-particle computation for a generalization of the classical problem of flat-plate constant-rate pitch and related motions, at frequencies and Reynolds numbers relevant to Micro Air Vehicle applications. The motivation is problems of maneuvering, perching and gust response. All of the examined flows evince a strong leading edge vortex. Increasing pitch rate tends to tighten the leading edge vortex and to produce a trailing-edge vortex system dominated by a counter-rotating pair. Pitch pivot point location is crucial to the leading edge vortex size and formation history, and to its subsequent behavior in convecting over the airfoil suction-side. Despite the respective limitations of the experiment and computations, agreement in vorticity fields between the two at an overlapping case at Re = 10,000 is good, whence it is possible to use the computation to obtain integrated force data unavailable in the experiment. These were studied for Re= 100 and 1000. Lift prediction from the computation shows a direct proportionality of lift to the pitch rate on the pitch upstroke. Finally, we compare pitch vs. plunge, and find that quasi-steady prediction is reasonably successful in predicting a combined pitch-plunge that effectively cancels the leading edge vortex, but not in canceling the trailing vortex system.

Human parsing is a challenging task because it is difficult to obtain accurate results of each part of human body. Precious Boltzmann Machine based methods reach good results on segmentation but are poor expression on human parts. In this paper, an approach is presented that exploits Shape Boltzmann Machine networks to improve the accuracy of human body parsing. The proposed Curve Correction method refines the final segmentation results. Experimental results show that the proposed method achieves good performance in body parsing, measured by Average Pixel Accuracy (aPA) against state-of-the-art methods on Penn-Fudan Pedestrians dataset and Pedestrian Parsing Surveillance Scenes dataset.

Considering their contribution to global warming, the sources and sinks of methane (CH4) should be accounted when undertaking a greenhouse gas inventory for grazed rangeland ecosystems. The aim of this study was to evaluate the mitigation potential of current ecological management programs implemented in the main rangeland regions of China. The influences of rangeland improvement, utilization and livestock production on CH4 flux/emission were assessed to estimate CH4 reduction potential. Results indicate that the grazed rangeland ecosystem is currently a net source of atmospheric CH4. However, there is potential to convert the ecosystem to a net sink by improving management practices. Previous assessments of capacity for CH4 uptake in grazed rangeland ecosystems have not considered improved livestock management practices and thus underestimated potential for CH4 uptake. Optimal fertilization, rest and light grazing, and intensification of livestock management contribute mitigation potential significantly.

This paper presents a novel approach to remove the specular reflections on the transparent plastic medicine package and automatically extract the randomly distributed pills inside. In this approach, three cameras are employed to take images of the package from different viewpoints. And these three images are used as input image set while the output is a series of small images of a single pill. And these images can be directly applied to the traditional single pill recognition algorithms. The experimental results show the reliability of our approach by measuring correct detection rate (100%), false detection rate (0%) and pill separation accuracy (98.4%). And the proposed method processes a set of three 725×725 sized images at 0.15s averagely on a Core i5-2400 3.1GHz PC.

we present a novel method to remove the specular reflections on the surface of transparent plastic film. Our approach uses four light sources with strategic positions to get four images. Based on the information that both reflection and shadow move a lot from image to image, we reconstruct a high quality image free from reflection and shadow by using a image set which is consist of four images.

A low-order point vortex model for the unsteady aerodynamics of agile ight of micro air vehicles is developed further in this work. A vortex is released from both the trailing and leading edges of the at plate section, and the strength of each is determined by enforcing the Kutta condition at the edges. We derive a new evolution equation for the vortex position by equating the rate of change of its impulse with that of an equivalent surrogate vortex with identical properties but constant strength. This approach leads to a model that admits more general criteria for ‘shedding’ (i.e. freezing the strength of the vortex) than the previous model developed by the authors, based on the Brown{Michael equation (AIAA Paper 2010-4281). We show that the results of the new model, when applied to a pitching or perching wing, agree much better with experiments and high-delity simulations than the previous model. Current limitations of the model and extensions to more general unsteady aerodynamic problems are discussed.

numerical simulations at Reynolds number 1000 and a low-order inviscid point vortex model. The pitching rate and axis position are systematically varied, and their eect on the generated aerodynamic forces is inspected. It is found that the maximum lift and drag developed during the pitch-up both increase nearly linearly with increasing pitch rate, though the rates of increase diminish as the pitching axis is moved aft. Furthermore, the maximum lift-to-drag ratio tends to saturate with increasing pitch rate, with the asymptotic value decreasing as the axis is moved aft. The forces predicted by the low-order inviscid Brown{Michael model are compared with the high-delity results. Good qualitative agreement is achieved, though the point vortex model tends to over-predict both components of force. The lift force obtained from the model is decomposed into inertial reaction and circulatory components, and their relative contributions are inspected.

Efficiency is an important issue in designing video architectures for action recognition. 3D CNNs have witnessed remarkable progress in action recognition from videos. However, compared with their 2D counterparts, 3D convolutions often introduce a large amount of parameters and cause high computational cost. To relieve this problem, we propose an efficient temporal module, termed as Temporal Enhancement-and-Interaction (TEI Module), which could be plugged into the existing 2D CNNs (denoted by TEINet). The TEI module presents a different paradigm to learn temporal features by decoupling the modeling of channel correlation and temporal interaction. First, it contains a Motion Enhanced Module (MEM) which is to enhance the motion-related features while suppress irrelevant information (e.g., background). Then, it introduces a Temporal Interaction Module (TIM) which supplements the temporal contextual information in a channel-wise manner. This two-stage modeling scheme is not only able to capture temporal structure flexibly and effectively, but also efficient for model inference. We conduct extensive experiments to verify the effectiveness of TEINet on several benchmarks (e.g., Something-Something V1&V2, Kinetics, UCF101 and HMDB51). Our proposed TEINet can achieve a good recognition accuracy on these datasets but still preserve a high efficiency., Accepted by AAAI 2020