Search

This paper reviews the NTIRE 2024 low light image enhancement challenge, highlighting the proposed solutions and results. The aim of this challenge is to discover an effective network design or solution capable of generating brighter, clearer, and visually appealing results when dealing with a variety of conditions, including ultra-high resolution (4K and beyond), non-uniform illumination, backlighting, extreme darkness, and night scenes. A notable total of 428 participants registered for the challenge, with 22 teams ultimately making valid submissions. This paper meticulously evaluates the state-of-the-art advancements in enhancing low-light images, reflecting the significant progress and creativity in this field., Comment: NTIRE 2024 Challenge Report

In real-world scenarios, images captured often suffer from blurring, noise, and other forms of image degradation, and due to sensor limitations, people usually can only obtain low dynamic range images. To achieve high-quality images, researchers have attempted various image restoration and enhancement operations on photographs, including denoising, deblurring, and high dynamic range imaging. However, merely performing a single type of image enhancement still cannot yield satisfactory images. In this paper, to deal with the challenge above, we propose the Composite Refinement Network (CRNet) to address this issue using multiple exposure images. By fully integrating information-rich multiple exposure inputs, CRNet can perform unified image restoration and enhancement. To improve the quality of image details, CRNet explicitly separates and strengthens high and low-frequency information through pooling layers, using specially designed Multi-Branch Blocks for effective fusion of these frequencies. To increase the receptive field and fully integrate input features, CRNet employs the High-Frequency Enhancement Module, which includes large kernel convolutions and an inverted bottleneck ConvFFN. Our model secured third place in the first track of the Bracketing Image Restoration and Enhancement Challenge, surpassing previous SOTA models in both testing metrics and visual quality., Comment: This paper is accepted by CVPR2024 Workshop, Code: https://github.com/CalvinYang0/CRNet

In real-world scenarios, due to a series of image degradations, obtaining high-quality, clear content photos is challenging. While significant progress has been made in synthesizing high-quality images, previous methods for image restoration and enhancement often overlooked the characteristics of different degradations. They applied the same structure to address various types of degradation, resulting in less-than-ideal restoration outcomes. Inspired by the notion that high/low frequency information is applicable to different degradations, we introduce HLNet, a Bracketing Image Restoration and Enhancement method based on high-low frequency decomposition. Specifically, we employ two modules for feature extraction: shared weight modules and non-shared weight modules. In the shared weight modules, we use SCConv to extract common features from different degradations. In the non-shared weight modules, we introduce the High-Low Frequency Decomposition Block (HLFDB), which employs different methods to handle high-low frequency information, enabling the model to address different degradations more effectively. Compared to other networks, our method takes into account the characteristics of different degradations, thus achieving higher-quality image restoration., Comment: This paper is accepted by CVPR 2024 Workshop, code: https://github.com/chengeng0613/HLNet

Weakly supervised video anomaly detection (WSVAD) is a challenging task. Generating fine-grained pseudo-labels based on weak-label and then self-training a classifier is currently a promising solution. However, since the existing methods use only RGB visual modality and the utilization of category text information is neglected, thus limiting the generation of more accurate pseudo-labels and affecting the performance of self-training. Inspired by the manual labeling process based on the event description, in this paper, we propose a novel pseudo-label generation and self-training framework based on Text Prompt with Normality Guidance (TPWNG) for WSVAD. Our idea is to transfer the rich language-visual knowledge of the contrastive language-image pre-training (CLIP) model for aligning the video event description text and corresponding video frames to generate pseudo-labels. Specifically, We first fine-tune the CLIP for domain adaptation by designing two ranking losses and a distributional inconsistency loss. Further, we propose a learnable text prompt mechanism with the assist of a normality visual prompt to further improve the matching accuracy of video event description text and video frames. Then, we design a pseudo-label generation module based on the normality guidance to infer reliable frame-level pseudo-labels. Finally, we introduce a temporal context self-adaptive learning module to learn the temporal dependencies of different video events more flexibly and accurately. Extensive experiments show that our method achieves state-of-the-art performance on two benchmark datasets, UCF-Crime and XD-Viole, Comment: Accepted to CVPR2024

Deep neural networks (DNNs) are known to be sensitive to adversarial input perturbations, leading to a reduction in either prediction accuracy or individual fairness. To jointly characterize the susceptibility of prediction accuracy and individual fairness to adversarial perturbations, we introduce a novel robustness definition termed robust accurate fairness. Informally, robust accurate fairness requires that predictions for an instance and its similar counterparts consistently align with the ground truth when subjected to input perturbations. We propose an adversarial attack approach dubbed RAFair to expose false or biased adversarial defects in DNN, which either deceive accuracy or compromise individual fairness. Then, we show that such adversarial instances can be effectively addressed by carefully designed benign perturbations, correcting their predictions to be accurate and fair. Our work explores the double-edged sword of input perturbations to robust accurate fairness in DNN and the potential of using benign perturbations to correct adversarial instances.

The low-rank quaternion matrix approximation has been successfully applied in many applications involving signal processing and color image processing. However, the cost of quaternion models for generating low-rank quaternion matrix approximation is sometimes considerable due to the computation of the quaternion singular value decomposition (QSVD), which limits their application to real large-scale data. To address this deficiency, an efficient quaternion matrix CUR (QMCUR) method for low-rank approximation is suggested, which provides significant acceleration in color image processing. We first explore the QMCUR approximation method, which uses actual columns and rows of the given quaternion matrix, instead of the costly QSVD. Additionally, two different sampling strategies are used to sample the above-selected columns and rows. Then, the perturbation analysis is performed on the QMCUR approximation of noisy versions of low-rank quaternion matrices. Extensive experiments on both synthetic and real data further reveal the superiority of the proposed algorithm compared with other algorithms for getting low-rank approximation, in terms of both efficiency and accuracy.

Understanding treatment heterogeneity is crucial for reliable decision-making in treatment evaluation and selection. While the conditional average treatment effect (CATE) is commonly used to capture treatment heterogeneity induced by covariates and design individualized treatment policies, it remains an averaging metric within subpopulations. This limitation prevents it from unveiling individual-level risks, potentially leading to misleading results. This article addresses this gap by examining individual risk for binary outcomes, specifically focusing on the fraction negatively affected (FNA) conditional on covariates -- a metric assessing the percentage of individuals experiencing worse outcomes with treatment compared to control. Under the strong ignorability assumption, FNA is unidentifiable, and we find that previous bounds are wide and practically unattainable except in certain degenerate cases. By introducing a plausible positive correlation assumption for the potential outcomes, we obtain significantly improved bounds compared to previous studies. We show that even with a positive and statistically significant CATE, the lower bound on FNA can be positive, i.e., in the best-case scenario many units will be harmed if receiving treatment. We establish a nonparametric sensitivity analysis framework for FNA using the Pearson correlation coefficient as the sensitivity parameter, thereby exploring the relationships among the correlation coefficient, FNA, and CATE. We also present a practical and tractable method for selecting the range of correlation coefficients. Furthermore, we propose flexible estimators for refined FNA bounds and prove their consistency and asymptotic normality.

Low-Light Image Enhancement (LLIE) task tends to restore the details and visual information from corrupted low-light images. Most existing methods learn the mapping function between low/normal-light images by Deep Neural Networks (DNNs) on sRGB and HSV color space. Nevertheless, enhancement involves amplifying image signals, and applying these color spaces to low-light images with a low signal-to-noise ratio can introduce sensitivity and instability into the enhancement process. Consequently, this results in the presence of color artifacts and brightness artifacts in the enhanced images. To alleviate this problem, we propose a novel trainable color space, named Horizontal/Vertical-Intensity (HVI). It not only decouples brightness and color from RGB channels to mitigate the instability during enhancement but also adapts to low-light images in different illumination ranges due to the trainable parameters. Further, we design a novel Color and Intensity Decoupling Network (CIDNet) with two branches dedicated to processing the decoupled image brightness and color in the HVI space. Within CIDNet, we introduce the Lightweight Cross-Attention (LCA) module to facilitate interaction between image structure and content information in both branches, while also suppressing noise in low-light images. Finally, we conducted 22 quantitative and qualitative experiments to show that the proposed CIDNet outperforms the state-of-the-art methods on 11 datasets. The code is available at https://github.com/Fediory/HVI-CIDNet., Comment: Qingsen Yan, Yixu Feng, Cheng Zhang contributed equally to this work. Corresponding author: Yanning Zhang

Video anomaly detection (VAD) with weak supervision has achieved remarkable performance in utilizing video-level labels to discriminate whether a video frame is normal or abnormal. However, current approaches are inherently limited to a closed-set setting and may struggle in open-world applications where there can be anomaly categories in the test data unseen during training. A few recent studies attempt to tackle a more realistic setting, open-set VAD, which aims to detect unseen anomalies given seen anomalies and normal videos. However, such a setting focuses on predicting frame anomaly scores, having no ability to recognize the specific categories of anomalies, despite the fact that this ability is essential for building more informed video surveillance systems. This paper takes a step further and explores open-vocabulary video anomaly detection (OVVAD), in which we aim to leverage pre-trained large models to detect and categorize seen and unseen anomalies. To this end, we propose a model that decouples OVVAD into two mutually complementary tasks -- class-agnostic detection and class-specific classification -- and jointly optimizes both tasks. Particularly, we devise a semantic knowledge injection module to introduce semantic knowledge from large language models for the detection task, and design a novel anomaly synthesis module to generate pseudo unseen anomaly videos with the help of large vision generation models for the classification task. These semantic knowledge and synthesis anomalies substantially extend our model's capability in detecting and categorizing a variety of seen and unseen anomalies. Extensive experiments on three widely-used benchmarks demonstrate our model achieves state-of-the-art performance on OVVAD task., Comment: Accepted to CVPR2024

There is growing interest in exploring causal effects in target populations via data combination. However, most approaches are tailored to specific settings and lack comprehensive comparative analyses. In this article, we focus on a typical scenario involving a source dataset and a target dataset. We first design six settings under covariate shift and conduct a comparative analysis by deriving the semiparametric efficiency bounds for the ATE in the target population. We then extend this analysis to six new settings that incorporate both covariate shift and posterior drift. Our study uncovers the key factors that influence efficiency gains and the ``effective sample size" when combining two datasets, with a particular emphasis on the roles of the variance ratio of potential outcomes between datasets and the derivatives of the posterior drift function. To the best of our knowledge, this is the first paper that explicitly explores the role of the posterior drift functions in causal inference. Additionally, we also propose novel methods for conducting sensitivity analysis to address violations of transportability between the two datasets. We empirically validate our findings by constructing locally efficient estimators and conducting extensive simulations. We demonstrate the proposed methods in two real-world applications.

Class imbalance exists in many classification problems, and since the data is designed for accuracy, imbalance in data classes can lead to classification challenges with a few classes having higher misclassification costs. The Backblaze dataset, a widely used dataset related to hard discs, has a small amount of failure data and a large amount of health data, which exhibits a serious class imbalance. This paper provides a comprehensive overview of research in the field of imbalanced data classification. The discussion is organized into three main aspects: data-level methods, algorithmic-level methods, and hybrid methods. For each type of method, we summarize and analyze the existing problems, algorithmic ideas, strengths, and weaknesses. Additionally, the challenges of unbalanced data classification are discussed, along with strategies to address them. It is convenient for researchers to choose the appropriate method according to their needs.

Large scale data storage is susceptible to failure. As disks are damaged and replaced, traditional machine learning models, which rely on historical data to make predictions, struggle to accurately predict disk failures. This paper presents a novel method for predicting disk failures by leveraging multi-layer domain adaptive learning techniques. First, disk data with numerous faults is selected as the source domain, and disk data with fewer faults is selected as the target domain. A training of the feature extraction network is performed with the selected origin and destination domains. The contrast between the two domains facilitates the transfer of diagnostic knowledge from the domain of source and target. According to the experimental findings, it has been demonstrated that the proposed technique can generate a reliable prediction model and improve the ability to predict failures on disk data with few failure samples.

Data class imbalance is a common problem in classification problems, where minority class samples are often more important and more costly to misclassify in a classification task. Therefore, it is very important to solve the data class imbalance classification problem. The SMART dataset exhibits an evident class imbalance, comprising a substantial quantity of healthy samples and a comparatively limited number of defective samples. This dataset serves as a reliable indicator of the disc's health status. In this paper, we obtain the best balanced disk SMART dataset for a specific classification model by mixing and integrating the data synthesised by multivariate generative adversarial networks (GAN) to balance the disk SMART dataset at the data level; and combine it with genetic algorithms to obtain higher disk fault classification prediction accuracy on a specific classification model.

To address the problem of insufficient failure data generated by disks and the imbalance between the number of normal and failure data. The existing Conditional Tabular Generative Adversarial Networks (CTGAN) deep learning methods have been proven to be effective in solving imbalance disk failure data. But CTGAN cannot learn the internal information of disk failure data very well. In this paper, a fault diagnosis method based on improved CTGAN, a classifier for specific category discrimination is added and a discriminator generate adversarial network based on residual network is proposed. We named it Residual Conditional Tabular Generative Adversarial Networks (RCTGAN). Firstly, to enhance the stability of system a residual network is utilized. RCTGAN uses a small amount of real failure data to synthesize fake fault data; Then, the synthesized data is mixed with the real data to balance the amount of normal and failure data; Finally, four classifier (multilayer perceptron, support vector machine, decision tree, random forest) models are trained using the balanced data set, and the performance of the models is evaluated using G-mean. The experimental results show that the data synthesized by the RCTGAN can further improve the fault diagnosis accuracy of the classifier.

In the domain of video surveillance, describing the behavior of each individual within the video is becoming increasingly essential, especially in complex scenarios with multiple individuals present. This is because describing each individual's behavior provides more detailed situational analysis, enabling accurate assessment and response to potential risks, ensuring the safety and harmony of public places. Currently, video-level captioning datasets cannot provide fine-grained descriptions for each individual's specific behavior. However, mere descriptions at the video-level fail to provide an in-depth interpretation of individual behaviors, making it challenging to accurately determine the specific identity of each individual. To address this challenge, we construct a human-centric video surveillance captioning dataset, which provides detailed descriptions of the dynamic behaviors of 7,820 individuals. Specifically, we have labeled several aspects of each person, such as location, clothing, and interactions with other elements in the scene, and these people are distributed across 1,012 videos. Based on this dataset, we can link individuals to their respective behaviors, allowing for further analysis of each person's behavior in surveillance videos. Besides the dataset, we propose a novel video captioning approach that can describe individual behavior in detail on a person-level basis, achieving state-of-the-art results. To facilitate further research in this field, we intend to release our dataset and code.

The recent contrastive language-image pre-training (CLIP) model has shown great success in a wide range of image-level tasks, revealing remarkable ability for learning powerful visual representations with rich semantics. An open and worthwhile problem is efficiently adapting such a strong model to the video domain and designing a robust video anomaly detector. In this work, we propose VadCLIP, a new paradigm for weakly supervised video anomaly detection (WSVAD) by leveraging the frozen CLIP model directly without any pre-training and fine-tuning process. Unlike current works that directly feed extracted features into the weakly supervised classifier for frame-level binary classification, VadCLIP makes full use of fine-grained associations between vision and language on the strength of CLIP and involves dual branch. One branch simply utilizes visual features for coarse-grained binary classification, while the other fully leverages the fine-grained language-image alignment. With the benefit of dual branch, VadCLIP achieves both coarse-grained and fine-grained video anomaly detection by transferring pre-trained knowledge from CLIP to WSVAD task. We conduct extensive experiments on two commonly-used benchmarks, demonstrating that VadCLIP achieves the best performance on both coarse-grained and fine-grained WSVAD, surpassing the state-of-the-art methods by a large margin. Specifically, VadCLIP achieves 84.51% AP and 88.02% AUC on XD-Violence and UCF-Crime, respectively. Code and features are released at https://github.com/nwpu-zxr/VadCLIP., Comment: Accept to AAAI2024

Global Navigation Satellite System (GNSS) is pervasive in navigation and positioning applications, where precise position and time referencing estimations are required. Conventional methods for GNSS positioning involve a two-step process, where intermediate measurements such as Doppler shift and time delay of received GNSS signals are computed and then used to solve for the receiver's position. Alternatively, Direct Position Estimation (DPE) was proposed to infer the position directly from the sampled signal without intermediate variables, yielding to superior levels of sensitivity and operation under challenging environments. However, the positioning resilience of DPE method is still under the threat of various interferences. Robust Interference Mitigation (RIM) processing has been studied and proved to be efficient against various interference in conventional two-step positioning (2SP) methods, and therefore worthy to be explored regarding its potential to enhance DPE. This article extends DPE methodology by incorporating RIM strategies that address the increasing need to protect GNSS receivers against intentional or unintentional interferences, such as jamming signals, which can deny GNSS-based positioning. RIM, which leverages robust statistics, was shown to provide competitive results in two-step approaches and is here employed in a high-sensitivity DPE framework with successful results. The article also provides a quantification of the loss of efficiency of using RIM when no interference is present and validates the proposed methodology on relevant interference cases, while the approach can be used to mitigate other common interference signals.