Yingli Tian

Ziyue Feng, Liang Yang, Longlong Jing et al.

Conventional self-supervised monocular depth prediction methods are based on a static environment assumption, which leads to accuracy degradation in dynamic scenes due to the mismatch and occlusion problems introduced by object motions. Existing dynamic-object-focused methods only partially solved the mismatch problem at the training loss level. In this paper, we accordingly propose a novel multi-frame monocular depth prediction method to solve these problems at both the prediction and supervision loss levels. Our method, called DynamicDepth, is a new framework trained via a self-supervised cycle consistent learning scheme. A Dynamic Object Motion Disentanglement (DOMD) module is proposed to disentangle object motions to solve the mismatch problem. Moreover, novel occlusion-aware Cost Volume and Re-projection Loss are designed to alleviate the occlusion effects of object motions. Extensive analyses and experiments on the Cityscapes and KITTI datasets show that our method significantly outperforms the state-of-the-art monocular depth prediction methods, especially in the areas of dynamic objects. Code is available at https://github.com/AutoAILab/DynamicDepth

Jiatong Li, Zheng Chen, Kai Liu et al.

This paper provides a review of the NTIRE 2026 challenge on mobile real-world image super-resolution, highlighting the proposed solutions and the resulting outcomes. The challenge aims to recover high-resolution (HR) images from low-resolution (LR) counterparts generated through unknown degradations with a x4 scaling factor while ensuring the models remain executable on mobile devices. The objective is to develop effective and efficient network designs or solutions that achieve state-of-the-art real-world image super-resolution performance. The track of the challenge evaluates performance using a weighted combination of image quality assessment (IQA) score and speedup ratios. The competition attracted 108 registrants, with 16 teams achieving a valid score in the final ranking. This collaborative effort advances the performance of mobile real-world image super-resolution while offering an in-depth overview of the latest trends in the field.

Zheng Chen, Kai Liu, Jingkai Wang et al.

This paper presents the NTIRE 2026 image super-resolution ($\times$4) challenge, one of the associated competitions of the NTIRE 2026 Workshop at CVPR 2026. The challenge aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs generated through bicubic downsampling with a $\times$4 scaling factor. The objective is to develop effective super-resolution solutions and analyze recent advances in the field. To reflect the evolving objectives of image super-resolution, the challenge includes two tracks: (1) a restoration track, which emphasizes pixel-wise fidelity and ranks submissions based on PSNR; and (2) a perceptual track, which focuses on visual realism and evaluates results using a perceptual score. A total of 194 participants registered for the challenge, with 31 teams submitting valid entries. This report summarizes the challenge design, datasets, evaluation protocol, main results, and methods of participating teams. The challenge provides a unified benchmark and offers insights into current progress and future directions in image super-resolution.

Haiyan Wang, Yingli Tian

Point cloud has drawn more and more research attention as well as real-world applications. However, many of these applications (e.g. autonomous driving and robotic manipulation) are actually based on sequential point clouds (i.e. four dimensions) because the information of the static point cloud data could provide is still limited. Recently, researchers put more and more effort into sequential point clouds. This paper presents an extensive review of the deep learning-based methods for sequential point cloud research including dynamic flow estimation, object detection \& tracking, point cloud segmentation, and point cloud forecasting. This paper further summarizes and compares the quantitative results of the reviewed methods over the public benchmark datasets. Finally, this paper is concluded by discussing the challenges in the current sequential point cloud research and pointing out insightful potential future research directions.

Saad Hassan, Matthew Seita, Larwan Berke et al.

We are releasing a dataset containing videos of both fluent and non-fluent signers using American Sign Language (ASL), which were collected using a Kinect v2 sensor. This dataset was collected as a part of a project to develop and evaluate computer vision algorithms to support new technologies for automatic detection of ASL fluency attributes. A total of 45 fluent and non-fluent participants were asked to perform signing homework assignments that are similar to the assignments used in introductory or intermediate level ASL courses. The data is annotated to identify several aspects of signing including grammatical features and non-manual markers. Sign language recognition is currently very data-driven and this dataset can support the design of recognition technologies, especially technologies that can benefit ASL learners. This dataset might also be interesting to ASL education researchers who want to contrast fluent and non-fluent signing.

Yuqi Li, Chuanguang Yang, Hansheng Zeng et al.

Spatiotemporal forecasting tasks, such as traffic flow, combustion dynamics, and weather forecasting, often require complex models that suffer from low training efficiency and high memory consumption. This paper proposes a lightweight framework, Spectral Decoupled Knowledge Distillation (termed SDKD), which transfers the multi-scale spatiotemporal representations from a complex teacher model to a more efficient lightweight student network. The teacher model follows an encoder-latent evolution-decoder architecture, where its latent evolution module decouples high-frequency details and low-frequency trends using convolution and Transformer (global low-frequency modeler). However, the multi-layer convolution and deconvolution structures result in slow training and high memory usage. To address these issues, we propose a frequency-aligned knowledge distillation strategy, which extracts multi-scale spectral features from the teacher's latent space, including both high and low frequency components, to guide the lightweight student model in capturing both local fine-grained variations and global evolution patterns. Experimental results show that SDKD significantly improves performance, achieving reductions of up to 81.3% in MSE and in MAE 52.3% on the Navier-Stokes equation dataset. The framework effectively captures both high-frequency variations and long-term trends while reducing computational complexity. Our codes are available at https://github.com/itsnotacie/SDKD

Yuqi Li, Kai Li, Xin Yin et al. · pku

Although deep learning has substantially advanced speech separation in recent years, most existing studies continue to prioritize separation quality while overlooking computational efficiency, an essential factor for low-latency speech processing in real-time applications. In this paper, we propose SepPrune, the first structured pruning framework specifically designed to compress deep speech separation models and reduce their computational cost. SepPrune begins by analyzing the computational structure of a given model to identify layers with the highest computational burden. It then introduces a differentiable masking strategy to enable gradient-driven channel selection. Based on the learned masks, SepPrune prunes redundant channels and fine-tunes the remaining parameters to recover performance. Extensive experiments demonstrate that this learnable pruning paradigm yields substantial advantages for channel pruning in speech separation models, outperforming existing methods. Notably, a model pruned with SepPrune can recover 85% of the performance of a pre-trained model (trained over hundreds of epochs) with only one epoch of fine-tuning, and achieves convergence 36$\times$ faster than training from scratch. Code is available at https://github.com/itsnotacie/SepPrune.

Yuqi Li, Qian Zhou, Huiran Duan et al.

Gait recognition is an attractive biometric modality for long-range and contact-free identification, but high-performing gait models often rely on deep and computationally expensive architectures that are difficult to deploy in practice. Knowledge distillation (KD) offers a natural way to transfer knowledge from a powerful teacher to an efficient student; however, standard KD is often less effective for part-structured gait models, where supervision is formed from both part-wise classification logits and part-wise retrieval embeddings. In this paper, we propose GaitKD, a distillation framework that decouples gait knowledge transfer into two complementary components: decision-level distillation and boundary-level distillation. Specifically, GaitKD aligns the teacher and student through part-calibrated logit distillation to transfer inter-class decision relations, while preserving the teacher-induced partitioning of the embedding space through an activation-boundary objective instead of direct feature regression. With a simple aligned part-wise design, GaitKD supports heterogeneous teacher-student gait models without introducing additional inference cost. Experimental results across multiple gait recognition benchmarks and teacher-student configurations show consistent improvements over strong gait baselines. Our study demonstrates that the two transfer components are complementary, and boundary-preserving distillation provides more stable performance than direct feature regression. Source code is available at https://github.com/liyiersan/GaitKD/

Elahe Vahdani, Yingli Tian

This paper tackles the challenge of point-supervised temporal action detection, wherein only a single frame is annotated for each action instance in the training set. Most of the current methods, hindered by the sparse nature of annotated points, struggle to effectively represent the continuous structure of actions or the inherent temporal and semantic dependencies within action instances. Consequently, these methods frequently learn merely the most distinctive segments of actions, leading to the creation of incomplete action proposals. This paper proposes POTLoc, a Pseudo-label Oriented Transformer for weakly-supervised Action Localization utilizing only point-level annotation. POTLoc is designed to identify and track continuous action structures via a self-training strategy. The base model begins by generating action proposals solely with point-level supervision. These proposals undergo refinement and regression to enhance the precision of the estimated action boundaries, which subsequently results in the production of `pseudo-labels' to serve as supplementary supervisory signals. The architecture of the model integrates a transformer with a temporal feature pyramid to capture video snippet dependencies and model actions of varying duration. The pseudo-labels, providing information about the coarse locations and boundaries of actions, assist in guiding the transformer for enhanced learning of action dynamics. POTLoc outperforms the state-of-the-art point-supervised methods on THUMOS'14 and ActivityNet-v1.2 datasets.

Huiran Duan, Qian Zhou, Zhongliang Guo et al.

Conventional gait de-identification methods often encounter an inherent trade-off: they either provide insufficient identity suppression or introduce spatiotemporal distortions that impede structure-sensitive downstream applications. We propose GaitProtector, an impersonation-driven gait de-identification framework that formulates privacy protection as a unified objective with two tightly coupled components: (i) obfuscation, which repels the protected gait from the source identity, and (ii) impersonation, which attracts it toward a selected target identity. The target identity serves as a semantic anchor that biases optimization toward structurally plausible gait patterns under the pretrained diffusion prior, helping preserve dominant body shape and motion dynamics. We instantiate this idea through a training-free diffusion latent optimization pipeline. Instead of retraining a generator for each dataset, we invert each input silhouette sequence into the latent trajectory of a pretrained 3D video diffusion model and iteratively optimize latent codes with a differentiable adversarial objective to synthesize protected gaits. Experiments on the CASIA-B dataset show that GaitProtector achieves a 56.7% impersonation success rate under black-box gait recognition and reduces Rank-1 identification accuracy from 89.6% to 15.0%, while maintaining favorable visual and temporal quality. We further evaluate downstream utility on the Scoliosis1K dataset, where diagnostic accuracy decreases only from 91.4% to 74.2%. To the best of our knowledge, this work is the first to leverage pretrained 3D diffusion priors in a training-free manner for silhouette-based gait de-identification.

Yuqi Li, Kuiye Ding, Chuanguang Yang et al.

Time Series foundation models (TSFMs) deliver strong forecasting performance through large-scale pretraining, but their large parameter sizes make deployment costly. While knowledge distillation offers a natural and effective approach for model compression, techniques developed for general machine learning tasks are not directly applicable to time series forecasting due to the unique characteristics. To address this, we present DistilTS, the first distillation framework specifically designed for TSFMs. DistilTS addresses two key challenges: (1) task difficulty discrepancy, specific to forecasting, where uniform weighting makes optimization dominated by easier short-term horizons, while long-term horizons receive weaker supervision; and (2) architecture discrepancy, a general challenge in distillation, for which we design an alignment mechanism in the time series forecasting. To overcome these issues, DistilTS introduces horizon-weighted objectives to balance learning across horizons, and a temporal alignment strategy that reduces architectural mismatch, enabling compact models. Experiments on multiple benchmarks demonstrate that DistilTS achieves forecasting performance comparable to full-sized TSFMs, while reducing parameters by up to 1/150 and accelerating inference by up to 6000x. Code is available at: https://github.com/itsnotacie/DistilTS-ICASSP2026.

Elahe Vahdani, Yingli Tian

This paper addresses the challenge of point-supervised temporal action detection, in which only one frame per action instance is annotated in the training set. Self-training aims to provide supplementary supervision for the training process by generating pseudo-labels (action proposals) from a base model. However, most current methods generate action proposals by applying manually designed thresholds to action classification probabilities and treating adjacent snippets as independent entities. As a result, these methods struggle to generate complete action proposals, exhibit sensitivity to fluctuations in action classification scores, and generate redundant and overlapping action proposals. This paper proposes a novel framework termed ADM-Loc, which stands for Actionness Distribution Modeling for point-supervised action Localization. ADM-Loc generates action proposals by fitting a composite distribution, comprising both Gaussian and uniform distributions, to the action classification signals. This fitting process is tailored to each action class present in the video and is applied separately for each action instance, ensuring the distinctiveness of their distributions. ADM-Loc significantly enhances the alignment between the generated action proposals and ground-truth action instances and offers high-quality pseudo-labels for self-training. Moreover, to model action boundary snippets, it enforces consistency in action classification scores during training by employing Gaussian kernels, supervised with the proposed loss functions. ADM-Loc outperforms the state-of-the-art point-supervised methods on THUMOS14 and ActivityNet-v1.2 datasets.

Yuqi Li, Junhao Dong, Chuanguang Yang et al.

Vision-Language Models (VLMs) are increasingly deployed in safety-critical applications, making their adversarial robustness a crucial concern. While adversarial knowledge distillation has shown promise in transferring robustness from teacher to student models, traditional single-teacher approaches suffer from limited knowledge diversity, slow convergence, and difficulty in balancing robustness and accuracy. To address these challenges, we propose MMT-ARD: a Multimodal Multi-Teacher Adversarial Robust Distillation framework. Our key innovation is a dual-teacher knowledge fusion architecture that collaboratively optimizes clean feature preservation and robust feature enhancement. To better handle challenging adversarial examples, we introduce a dynamic weight allocation strategy based on teacher confidence, enabling adaptive focus on harder samples. Moreover, to mitigate bias among teachers, we design an adaptive sigmoid-based weighting function that balances the strength of knowledge transfer across modalities. Extensive experiments on ImageNet and zero-shot benchmarks demonstrate that MMT-ARD improves robust accuracy by +4.32% and zero-shot accuracy by +3.5% on the ViT-B-32 model, while achieving a 2.3x increase in training efficiency over traditional single-teacher methods. These results highlight the effectiveness and scalability of MMT-ARD in enhancing the adversarial robustness of multimodal large models. Our codes are available at https://github.com/itsnotacie/MMT-ARD.

Shaoheng Wang, Yao Lu, Yuqi Li et al.

As a parameter efficient fine-tuning (PEFT) method, low-rank adaptation (LoRA) can save significant costs in storage and computing, but its strong adaptability to a single task is often accompanied by insufficient cross-task generalization capabilities. To improve this, existing work combines LoRA with mixture-of-experts (MoE) to enhance the model's adaptability through expert modules and routing mechanisms. However, existing LoRA-MoE methods lack unified standards in models, datasets, hyperparameters, and evaluation methods, making it difficult to conduct fair comparisons between different methods. To this end, we proposed a unified benchmark named LoRALib. Specifically, we standardized datasets from $40$ downstream tasks into a unified format, fine-tuned them using the same hyperparameters and obtained $680$ LoRA modules across $17$ model architectures. Based on this LoRA library, we conduct large-scale experiments on $3$ representative LoRA-MoE methods and different LoRA selection mechanisms using the open-sourced testing tool OpenCompass. Extensive experiments show that LoRAMoE performs best, and that prioritizing LoRAs relevant to the target task can further improve the performance of MoE. We hope these findings will inspire future work. Our datasets and LoRA library are available at https://huggingface.co/datasets/YaoLuzjut/LoRAOcean_dataset and https://huggingface.co/YaoLuzjut/models.

Ziyue Feng, Longlong Jing, Peng Yin et al.

Self-supervised monocular depth prediction provides a cost-effective solution to obtain the 3D location of each pixel. However, the existing approaches usually lead to unsatisfactory accuracy, which is critical for autonomous robots. In this paper, we propose FusionDepth, a novel two-stage network to advance the self-supervised monocular dense depth learning by leveraging low-cost sparse (e.g. 4-beam) LiDAR. Unlike the existing methods that use sparse LiDAR mainly in a manner of time-consuming iterative post-processing, our model fuses monocular image features and sparse LiDAR features to predict initial depth maps. Then, an efficient feed-forward refine network is further designed to correct the errors in these initial depth maps in pseudo-3D space with real-time performance. Extensive experiments show that our proposed model significantly outperforms all the state-of-the-art self-supervised methods, as well as the sparse-LiDAR-based methods on both self-supervised monocular depth prediction and completion tasks. With the accurate dense depth prediction, our model outperforms the state-of-the-art sparse-LiDAR-based method (Pseudo-LiDAR++) by more than 68% for the downstream task monocular 3D object detection on the KITTI Leaderboard. Code is available at https://github.com/AutoAILab/FusionDepth

Hao Wu, Yuqi Li, Yuan Gao et al.

Existing robot video world models are typically trained with low-level objectives such as reconstruction and perceptual similarity, which are poorly aligned with the capabilities that matter most for robot decision making, including instruction following, manipulation success, and physical plausibility. They also suffer from error accumulation in long-horizon autoregressive prediction. We present RoboAlign-R1, a framework that combines reward-aligned post-training with stabilized long-horizon inference for robot video world models. We construct RobotWorldBench, a benchmark of 10,000 annotated video-instruction pairs collected from four robot data sources, and train a multimodal teacher judge, RoboAlign-Judge, to provide fine-grained six-dimensional evaluation of generated videos. We then distill the teacher into a lightweight student reward model for efficient reinforcement-learning-based post-training. To reduce long-horizon rollout drift, we further introduce Sliding Window Re-encoding (SWR), a training-free inference strategy that periodically refreshes the generation context. Under our in-domain evaluation protocol, RoboAlign-R1 improves the aggregate six-dimension score by 10.1% over the strongest baseline, including gains of 7.5% on Manipulation Accuracy and 4.6% on Instruction Following; these ranking improvements are further supported by an external VLM-based cross-check and a blinded human study. Meanwhile, SWR improves long-horizon prediction quality with only about 1% additional latency, yielding a 2.8% gain in SSIM and a 9.8% reduction in LPIPS. Together, these results show that reward-aligned post-training and stabilized long-horizon decoding improve task consistency, physical realism, and long-horizon prediction quality in robot video world models.

Yuqi Li, Yao Lu, Junhao Dong et al.

Layer pruning has emerged as a potent approach to remove redundant layers in the pre-trained network on the purpose of reducing network size and improve computational efficiency. However, existing layer pruning methods mostly overlook the intrinsic connections and inter-dependencies between different layers within complicated deep neural networks. This oversight can result in pruned models that do not preserve the essential characteristics of the pre-trained network as effectively as desired. To address these limitations, we propose a Similarity-Guided Layer Partition (SGLP) Pruning, a novel pruning framework that exploits representation similarity to guide efficient and informed layer removal for compressing large deep models. Our method begins by employing Centered Kernel Alignment (CKA) to quantify representational similarity between layers, uncovering structural patterns within the network. We then apply Fisher Optimal Segmentation on the similarity matrix to partition the network into semantically coherent layer segments. This segmentation allows pruning decisions to respect layer interdependencies and preserve essential knowledge. Within each segment, we introduce a fine-tuning-free importance evaluation using GradNorm, identifying and removing redundant layers in a targeted, segment-wise manner. Experimental results on both image classification tasks and large language models (LLMs) demonstrate that our proposed SGLP outperforms the state-of-the-art methods in accuracy and efficiency. Our approach achieves significant model compression with minimal performance degradation, making it well-suited for deployment in resource-limited environments.

Yuqi Li, Chuanguang Yang, Junhao Dong et al.

The success of large-scale visual language pretraining (VLP) models has driven widespread adoption of image-text retrieval tasks. However, their deployment on mobile devices remains limited due to large model sizes and computational complexity. We propose Adaptive Multi-Modal Multi-Teacher Knowledge Distillation (AMMKD), a novel framework that integrates multi-modal feature fusion, multi-teacher distillation, and adaptive optimization to deliver lightweight yet effective retrieval models. Specifically, our method begins with a feature fusion network that extracts and merges discriminative features from both the image and text modalities. To reduce model parameters and further improve performance, we design a multi-teacher knowledge distillation framework to pre-train two CLIP teacher models. We decouple modalities by pre-computing and storing text features as class vectors via the teacher text encoder to enhance efficiency. To better align teacher and student outputs, we apply KL scatter for probability distribution matching. Finally, we design an adaptive dynamic weighting scheme that treats multi-teacher distillation as a multi-objective optimization problem. By leveraging gradient space diversity, we dynamically adjust the influence of each teacher, reducing conflicts and guiding the student toward more optimal learning directions. Extensive experiments on three benchmark datasets demonstrate that AMMKD achieves superior performance while significantly reducing model complexity, validating its effectiveness and flexibility.

Ali Salar, Qing Liu, Yingli Tian et al.

The rapid growth of social media has led to the widespread sharing of individual portrait images, which pose serious privacy risks due to the capabilities of automatic face recognition (AFR) systems for mass surveillance. Hence, protecting facial privacy against unauthorized AFR systems is essential. Inspired by the generation capability of the emerging diffusion models, recent methods employ diffusion models to generate adversarial face images for privacy protection. However, they suffer from the diffusion purification effect, leading to a low protection success rate (PSR). In this paper, we first propose learning unconditional embeddings to increase the learning capacity for adversarial modifications and then use them to guide the modification of the adversarial latent code to weaken the diffusion purification effect. Moreover, we integrate an identity-preserving structure to maintain structural consistency between the original and generated images, allowing human observers to recognize the generated image as having the same identity as the original. Extensive experiments conducted on two public datasets, i.e., CelebA-HQ and LADN, demonstrate the superiority of our approach. The protected faces generated by our method outperform those produced by existing facial privacy protection approaches in terms of transferability and natural appearance.

Weilun Feng, Chuanguang Yang, Haotong Qin et al.

Diffusion models have demonstrated remarkable performance on vision generation tasks. However, the high computational complexity hinders its wide application on edge devices. Quantization has emerged as a promising technique for inference acceleration and memory reduction. However, existing quantization methods do not generalize well under extremely low-bit (2-4 bit) quantization. Directly applying these methods will cause severe performance degradation. We identify that the existing quantization framework suffers from the outlier-unfriendly quantizer design, suboptimal initialization, and optimization strategy. We present MPQ-DMv2, an improved \textbf{M}ixed \textbf{P}recision \textbf{Q}uantization framework for extremely low-bit \textbf{D}iffusion \textbf{M}odels. For the quantization perspective, the imbalanced distribution caused by salient outliers is quantization-unfriendly for uniform quantizer. We propose \textit{Flexible Z-Order Residual Mixed Quantization} that utilizes an efficient binary residual branch for flexible quant steps to handle salient error. For the optimization framework, we theoretically analyzed the convergence and optimality of the LoRA module and propose \textit{Object-Oriented Low-Rank Initialization} to use prior quantization error for informative initialization. We then propose \textit{Memory-based Temporal Relation Distillation} to construct an online time-aware pixel queue for long-term denoising temporal information distillation, which ensures the overall temporal consistency between quantized and full-precision model. Comprehensive experiments on various generation tasks show that our MPQ-DMv2 surpasses current SOTA methods by a great margin on different architectures, especially under extremely low-bit widths.

Haiyan Wang, Will Hutchcroft, Yuguang Li et al.

In this paper, we propose a new deep learning-based method for estimating room layout given a pair of 360 panoramas. Our system, called Position-aware Stereo Merging Network or PSMNet, is an end-to-end joint layout-pose estimator. PSMNet consists of a Stereo Pano Pose (SP2) transformer and a novel Cross-Perspective Projection (CP2) layer. The stereo-view SP2 transformer is used to implicitly infer correspondences between views, and can handle noisy poses. The pose-aware CP2 layer is designed to render features from the adjacent view to the anchor (reference) view, in order to perform view fusion and estimate the visible layout. Our experiments and analysis validate our method, which significantly outperforms the state-of-the-art layout estimators, especially for large and complex room spaces.

Kien Nguyen, Clinton Fookes, Sridha Sridharan et al.

The rapid emergence of airborne platforms and imaging sensors are enabling new forms of aerial surveillance due to their unprecedented advantages in scale, mobility, deployment and covert observation capabilities. This paper provides a comprehensive overview of human-centric aerial surveillance tasks from a computer vision and pattern recognition perspective. It aims to provide readers with an in-depth systematic review and technical analysis of the current state of aerial surveillance tasks using drones, UAVs and other airborne platforms. The main object of interest is humans, where single or multiple subjects are to be detected, identified, tracked, re-identified and have their behavior analyzed. More specifically, for each of these four tasks, we first discuss unique challenges in performing these tasks in an aerial setting compared to a ground-based setting. We then review and analyze the aerial datasets publicly available for each task, and delve deep into the approaches in the aerial literature and investigate how they presently address the aerial challenges. We conclude the paper with discussion on the missing gaps and open research questions to inform future research avenues.

Zhimin Chen, Longlong Jing, Yang Liang et al.

In recent years, semi-supervised learning has been widely explored and shows excellent data efficiency for 2D data. There is an emerging need to improve data efficiency for 3D tasks due to the scarcity of labeled 3D data. This paper explores how the coherence of different modelities of 3D data (e.g. point cloud, image, and mesh) can be used to improve data efficiency for both 3D classification and retrieval tasks. We propose a novel multimodal semi-supervised learning framework by introducing instance-level consistency constraint and a novel multimodal contrastive prototype (M2CP) loss. The instance-level consistency enforces the network to generate consistent representations for multimodal data of the same object regardless of its modality. The M2CP maintains a multimodal prototype for each class and learns features with small intra-class variations by minimizing the feature distance of each object to its prototype while maximizing the distance to the others. Our proposed framework significantly outperforms all the state-of-the-art counterparts for both classification and retrieval tasks by a large margin on the modelNet10 and ModelNet40 datasets.

Elahe Vahdani, Yingli Tian

Understanding human behavior and activity facilitates advancement of numerous real-world applications, and is critical for video analysis. Despite the progress of action recognition algorithms in trimmed videos, the majority of real-world videos are lengthy and untrimmed with sparse segments of interest. The task of temporal activity detection in untrimmed videos aims to localize the temporal boundary of actions and classify the action categories. Temporal activity detection task has been investigated in full and limited supervision settings depending on the availability of action annotations. This paper provides an extensive overview of deep learning-based algorithms to tackle temporal action detection in untrimmed videos with different supervision levels including fully-supervised, weakly-supervised, unsupervised, self-supervised, and semi-supervised. In addition, this paper also reviews advances in spatio-temporal action detection where actions are localized in both temporal and spatial dimensions. Moreover, the commonly used action detection benchmark datasets and evaluation metrics are described, and the performance of the state-of-the-art methods are compared. Finally, real-world applications of temporal action detection in untrimmed videos and a set of future directions are discussed.

Haiyan Wang, Jiahao Pang, Muhammad A. Lodhi et al.

Scene flow depicts the dynamics of a 3D scene, which is critical for various applications such as autonomous driving, robot navigation, AR/VR, etc. Conventionally, scene flow is estimated from dense/regular RGB video frames. With the development of depth-sensing technologies, precise 3D measurements are available via point clouds which have sparked new research in 3D scene flow. Nevertheless, it remains challenging to extract scene flow from point clouds due to the sparsity and irregularity in typical point cloud sampling patterns. One major issue related to irregular sampling is identified as the randomness during point set abstraction/feature extraction -- an elementary process in many flow estimation scenarios. A novel Spatial Abstraction with Attention (SA^2) layer is accordingly proposed to alleviate the unstable abstraction problem. Moreover, a Temporal Abstraction with Attention (TA^2) layer is proposed to rectify attention in temporal domain, leading to benefits with motions scaled in a larger range. Extensive analysis and experiments verified the motivation and significant performance gains of our method, dubbed as Flow Estimation via Spatial-Temporal Attention (FESTA), when compared to several state-of-the-art benchmarks of scene flow estimation.

Longlong Jing, Elahe Vahdani, Jiaxing Tan et al.

Cross-modal retrieval aims to learn discriminative and modal-invariant features for data from different modalities. Unlike the existing methods which usually learn from the features extracted by offline networks, in this paper, we propose an approach to jointly train the components of cross-modal retrieval framework with metadata, and enable the network to find optimal features. The proposed end-to-end framework is updated with three loss functions: 1) a novel cross-modal center loss to eliminate cross-modal discrepancy, 2) cross-entropy loss to maximize inter-class variations, and 3) mean-square-error loss to reduce modality variations. In particular, our proposed cross-modal center loss minimizes the distances of features from objects belonging to the same class across all modalities. Extensive experiments have been conducted on the retrieval tasks across multi-modalities, including 2D image, 3D point cloud, and mesh data. The proposed framework significantly outperforms the state-of-the-art methods on the ModelNet40 dataset.

Yucheng Chen, Yingli Tian, Mingyi He

Vision-based monocular human pose estimation, as one of the most fundamental and challenging problems in computer vision, aims to obtain posture of the human body from input images or video sequences. The recent developments of deep learning techniques have been brought significant progress and remarkable breakthroughs in the field of human pose estimation. This survey extensively reviews the recent deep learning-based 2D and 3D human pose estimation methods published since 2014. This paper summarizes the challenges, main frameworks, benchmark datasets, evaluation metrics, performance comparison, and discusses some promising future research directions.

Longlong Jing, Yucheng Chen, Ling Zhang et al.

Most of the existing self-supervised feature learning methods for 3D data either learn 3D features from point cloud data or from multi-view images. By exploring the inherent multi-modality attributes of 3D objects, in this paper, we propose to jointly learn modal-invariant and view-invariant features from different modalities including image, point cloud, and mesh with heterogeneous networks for 3D data. In order to learn modal- and view-invariant features, we propose two types of constraints: cross-modal invariance constraint and cross-view invariant constraint. Cross-modal invariance constraint forces the network to maximum the agreement of features from different modalities for same objects, while the cross-view invariance constraint forces the network to maximum agreement of features from different views of images for same objects. The quality of learned features has been tested on different downstream tasks with three modalities of data including point cloud, multi-view images, and mesh. Furthermore, the invariance cross different modalities and views are evaluated with the cross-modal retrieval task. Extensive evaluation results demonstrate that the learned features are robust and have strong generalizability across different tasks.

Elahe Vahdani, Longlong Jing, Yingli Tian et al.

As part of the development of an educational tool that can help students achieve fluency in American Sign Language (ASL) through independent and interactive practice with immediate feedback, this paper introduces a near real-time system to recognize grammatical errors in continuous signing videos without necessarily identifying the entire sequence of signs. Our system automatically recognizes if performance of ASL sentences contains grammatical errors made by ASL students. We first recognize the ASL grammatical elements including both manual gestures and nonmanual signals independently from multiple modalities (i.e. hand gestures, facial expressions, and head movements) by 3D-ResNet networks. Then the temporal boundaries of grammatical elements from different modalities are examined to detect ASL grammatical mistakes by using a sliding window-based approach. We have collected a dataset of continuous sign language, ASL-HW-RGBD, covering different aspects of ASL grammars for training and testing. Our system is able to recognize grammatical elements on ASL-HW-RGBD from manual gestures, facial expressions, and head movements and successfully detect 8 ASL grammatical mistakes.

Haiyan Wang, Xuejian Rong, Liang Yang et al.

The deficiency of 3D segmentation labels is one of the main obstacles to effective point cloud segmentation, especially for scenes in the wild with varieties of different objects. To alleviate this issue, we propose a novel deep graph convolutional network-based framework for large-scale semantic scene segmentation in point clouds with sole 2D supervision. Different with numerous preceding multi-view supervised approaches focusing on single object point clouds, we argue that 2D supervision is capable of providing sufficient guidance information for training 3D semantic segmentation models of natural scene point clouds while not explicitly capturing their inherent structures, even with only single view per training sample. Specifically, a Graph-based Pyramid Feature Network (GPFN) is designed to implicitly infer both global and local features of point sets and an Observability Network (OBSNet) is introduced to further solve object occlusion problem caused by complicated spatial relations of objects in 3D scenes. During the projection process, perspective rendering and semantic fusion modules are proposed to provide refined 2D supervision signals for training along with a 2D-3D joint optimization strategy. Extensive experimental results demonstrate the effectiveness of our 2D supervised framework, which achieves comparable results with the state-of-the-art approaches trained with full 3D labels, for semantic point cloud segmentation on the popular SUNCG synthetic dataset and S3DIS real-world dataset.

Longlong Jing, Yucheng Chen, Ling Zhang et al.

The success of supervised learning requires large-scale ground truth labels which are very expensive, time-consuming, or may need special skills to annotate. To address this issue, many self- or un-supervised methods are developed. Unlike most existing self-supervised methods to learn only 2D image features or only 3D point cloud features, this paper presents a novel and effective self-supervised learning approach to jointly learn both 2D image features and 3D point cloud features by exploiting cross-modality and cross-view correspondences without using any human annotated labels. Specifically, 2D image features of rendered images from different views are extracted by a 2D convolutional neural network, and 3D point cloud features are extracted by a graph convolution neural network. Two types of features are fed into a two-layer fully connected neural network to estimate the cross-modality correspondence. The three networks are jointly trained (i.e. cross-modality) by verifying whether two sampled data of different modalities belong to the same object, meanwhile, the 2D convolutional neural network is additionally optimized through minimizing intra-object distance while maximizing inter-object distance of rendered images in different views (i.e. cross-view). The effectiveness of the learned 2D and 3D features is evaluated by transferring them on five different tasks including multi-view 2D shape recognition, 3D shape recognition, multi-view 2D shape retrieval, 3D shape retrieval, and 3D part-segmentation. Extensive evaluations on all the five different tasks across different datasets demonstrate strong generalization and effectiveness of the learned 2D and 3D features by the proposed self-supervised method.

Longlong Jing, Toufiq Parag, Zhe Wu et al.

We propose a semi-supervised learning approach for video classification, VideoSSL, using convolutional neural networks (CNN). Like other computer vision tasks, existing supervised video classification methods demand a large amount of labeled data to attain good performance. However, annotation of a large dataset is expensive and time consuming. To minimize the dependence on a large annotated dataset, our proposed semi-supervised method trains from a small number of labeled examples and exploits two regulatory signals from unlabeled data. The first signal is the pseudo-labels of unlabeled examples computed from the confidences of the CNN being trained. The other is the normalized probabilities, as predicted by an image classifier CNN, that captures the information about appearances of the interesting objects in the video. We show that, under the supervision of these guiding signals from unlabeled examples, a video classification CNN can achieve impressive performances utilizing a small fraction of annotated examples on three publicly available datasets: UCF101, HMDB51 and Kinetics.

Jingya Liu, Liangliang Cao, Oguz Akin et al.

Accurate detection of pulmonary nodules with high sensitivity and specificity is essential for automatic lung cancer diagnosis from CT scans. Although many deep learning-based algorithms make great progress for improving the accuracy of nodule detection, the high false positive rate is still a challenging problem which limits the automatic diagnosis in routine clinical practice. Moreover, the CT scans collected from multiple manufacturers may affect the robustness of Computer-aided diagnosis (CAD) due to the differences in intensity scales and machine noises. In this paper, we propose a novel self-supervised learning assisted pulmonary nodule detection framework based on a 3D Feature Pyramid Network (3DFPN) to improve the sensitivity of nodule detection by employing multi-scale features to increase the resolution of nodules, as well as a parallel top-down path to transit the high-level semantic features to complement low-level general features. Furthermore, a High Sensitivity and Specificity (HS2) network is introduced to eliminate the false positive nodule candidates by tracking the appearance changes in continuous CT slices of each nodule candidate on Location History Images (LHI). In addition, in order to improve the performance consistency of the proposed framework across data captured by different CT scanners without using additional annotations, an effective self-supervised learning schema is applied to learn spatiotemporal features of CT scans from large-scale unlabeled data. The performance and robustness of our method are evaluated on several publicly available datasets with significant performance improvements. The proposed framework is able to accurately detect pulmonary nodules with high sensitivity and specificity and achieves 90.6% sensitivity with 1/8 false positive per scan which outperforms the state-of-the-art results 15.8% on LUNA16 dataset.

Jingya Liu, Liangliang Cao, Oguz Akin et al.

Accurate detection of pulmonary nodules with high sensitivity and specificity is essential for automatic lung cancer diagnosis from CT scans. Although many deep learning-based algorithms make great progress for improving the accuracy of nodule detection, the high false positive rate is still a challenging problem which limited the automatic diagnosis in routine clinical practice. In this paper, we propose a novel pulmonary nodule detection framework based on a 3D Feature Pyramid Network (3DFPN) to improve the sensitivity of nodule detection by employing multi-scale features to increase the resolution of nodules, as well as a parallel top-down path to transit the high-level semantic features to complement low-level general features. Furthermore, a High Sensitivity and Specificity (HS$^2$) network is introduced to eliminate the falsely detected nodule candidates by tracking the appearance changes in continuous CT slices of each nodule candidate. The proposed framework is evaluated on the public Lung Nodule Analysis (LUNA16) challenge dataset. Our method is able to accurately detect lung nodules at high sensitivity and specificity and achieves $90.4\%$ sensitivity with 1/8 false positive per scan which outperforms the state-of-the-art results $15.6\%$.

Longlong Jing, Elahe Vahdani, Matt Huenerfauth et al.

In this paper, we propose a 3D Convolutional Neural Network (3DCNN) based multi-stream framework to recognize American Sign Language (ASL) manual signs (consisting of movements of the hands, as well as non-manual face movements in some cases) in real-time from RGB-D videos, by fusing multimodality features including hand gestures, facial expressions, and body poses from multi-channels (RGB, depth, motion, and skeleton joints). To learn the overall temporal dynamics in a video, a proxy video is generated by selecting a subset of frames for each video which are then used to train the proposed 3DCNN model. We collect a new ASL dataset, ASL-100-RGBD, which contains 42 RGB-D videos captured by a Microsoft Kinect V2 camera, each of 100 ASL manual signs, including RGB channel, depth maps, skeleton joints, face features, and HDface. The dataset is fully annotated for each semantic region (i.e. the time duration of each word that the human signer performs). Our proposed method achieves 92.88 accuracy for recognizing 100 ASL words in our newly collected ASL-100-RGBD dataset. The effectiveness of our framework for recognizing hand gestures from RGB-D videos is further demonstrated on the Chalearn IsoGD dataset and achieves 76 accuracy which is 5.51 higher than the state-of-the-art work in terms of average fusion by using only 5 channels instead of 12 channels in the previous work.

Shuangyue Yu, Tzu-Hao Huang, Dianpeng Wang et al.

This paper presents design and control innovations of wearable robots that tackle two barriers to widespread adoption of powered exoskeletons, namely restriction of human movement and versatile control of wearable co-robot systems. First, the proposed quasi-direct drive actuation comprising of our customized high torque density motors and low ratio transmission mechanism significantly reduces the mass of the robot and produces high backdrivability. Second, we derive a biomechanics model-based control that generates biological torque profile for versatile control of both squat and stoop lifting assistance. The control algorithm detects lifting postures using compact inertial measurement unit (IMU) sensors to generate an assistive profile that is proportional to the biological torque produced from our model. Experimental results demonstrate that the robot exhibits low mechanical impedance (1.5 Nm resistive torque) when it is unpowered and 0.5 Nm resistive torque with zero-torque tracking control. Root mean square (RMS) error of torque tracking is less than 0.29 Nm (1.21% error of 24 Nm peak torque). Compared with squatting without the exoskeleton, the controller reduces 87.5%, 80% and 75% of the of three knee extensor muscles (average peak EMG of 3 healthy subjects) during squat with 50% of biological torque assistance.

Longlong Jing, Yingli Tian

Large-scale labeled data are generally required to train deep neural networks in order to obtain better performance in visual feature learning from images or videos for computer vision applications. To avoid extensive cost of collecting and annotating large-scale datasets, as a subset of unsupervised learning methods, self-supervised learning methods are proposed to learn general image and video features from large-scale unlabeled data without using any human-annotated labels. This paper provides an extensive review of deep learning-based self-supervised general visual feature learning methods from images or videos. First, the motivation, general pipeline, and terminologies of this field are described. Then the common deep neural network architectures that used for self-supervised learning are summarized. Next, the main components and evaluation metrics of self-supervised learning methods are reviewed followed by the commonly used image and video datasets and the existing self-supervised visual feature learning methods. Finally, quantitative performance comparisons of the reviewed methods on benchmark datasets are summarized and discussed for both image and video feature learning. At last, this paper is concluded and lists a set of promising future directions for self-supervised visual feature learning.

Jiaxing Tan, Longlong Jing, Yumei Huo et al.

Lung segmentation in computerized tomography (CT) images is an important procedure in various lung disease diagnosis. Most of the current lung segmentation approaches are performed through a series of procedures with manually empirical parameter adjustments in each step. Pursuing an automatic segmentation method with fewer steps, in this paper, we propose a novel deep learning Generative Adversarial Network (GAN) based lung segmentation schema, which we denote as LGAN. Our proposed schema can be generalized to different kinds of neural networks for lung segmentation in CT images and is evaluated on a dataset containing 220 individual CT scans with two metrics: segmentation quality and shape similarity. Also, we compared our work with current state of the art methods. The results obtained with this study demonstrate that the proposed LGAN schema can be used as a promising tool for automatic lung segmentation due to its simplified procedure as well as its good performance.

Longlong Jing, Yucheng Chen, Yingli Tian

Deep neural network-based semantic segmentation generally requires large-scale cost extensive annotations for training to obtain better performance. To avoid pixel-wise segmentation annotations which are needed for most methods, recently some researchers attempted to use object-level labels (e.g. bounding boxes) or image-level labels (e.g. image categories). In this paper, we propose a novel recursive coarse-to-fine semantic segmentation framework based on only image-level category labels. For each image, an initial coarse mask is first generated by a convolutional neural network-based unsupervised foreground segmentation model and then is enhanced by a graph model. The enhanced coarse mask is fed to a fully convolutional neural network to be recursively refined. Unlike existing image-level label-based semantic segmentation methods which require to label all categories for images contain multiple types of objects, our framework only needs one label for each image and can handle images contains multi-category objects. With only trained on ImageNet, our framework achieves comparable performance on PASCAL VOC dataset as other image-level label-based state-of-the-arts of semantic segmentation. Furthermore, our framework can be easily extended to foreground object segmentation task and achieves comparable performance with the state-of-the-art supervised methods on the Internet Object dataset.

Benjamin Planche, Xuejian Rong, Ziyan Wu et al.

We present a method to incrementally generate complete 2D or 3D scenes with the following properties: (a) it is globally consistent at each step according to a learned scene prior, (b) real observations of a scene can be incorporated while observing global consistency, (c) unobserved regions can be hallucinated locally in consistence with previous observations, hallucinations and global priors, and (d) hallucinations are statistical in nature, i.e., different scenes can be generated from the same observations. To achieve this, we model the virtual scene, where an active agent at each step can either perceive an observed part of the scene or generate a local hallucination. The latter can be interpreted as the agent's expectation at this step through the scene and can be applied to autonomous navigation. In the limit of observing real data at each point, our method converges to solving the SLAM problem. It can otherwise sample entirely imagined scenes from prior distributions. Besides autonomous agents, applications include problems where large data is required for building robust real-world applications, but few samples are available. We demonstrate efficacy on various 2D as well as 3D data.

Yuancheng Ye, Xiaodong Yang, Yingli Tian

In this paper, we address the challenging problem of spatial and temporal action detection in videos. We first develop an effective approach to localize frame-level action regions through integrating static and kinematic information by the early- and late-fusion detection scheme. With the intention of exploring important temporal connections among the detected action regions, we propose a tracking-by-point-matching algorithm to stitch the discrete action regions into a continuous spatio-temporal action tube. Recurrent 3D convolutional neural network is used to predict action categories and determine temporal boundaries of the generated tubes. We then introduce an action footprint map to refine the candidate tubes based on the action-specific spatial characteristics preserved in the convolutional layers of R3DCNN. In the extensive experiments, our method achieves superior detection results on the three public benchmark datasets: UCFSports, J-HMDB and UCF101.

Longlong Jing, Xiaodong Yang, Jingen Liu et al.

The success of deep neural networks generally requires a vast amount of training data to be labeled, which is expensive and unfeasible in scale, especially for video collections. To alleviate this problem, in this paper, we propose 3DRotNet: a fully self-supervised approach to learn spatiotemporal features from unlabeled videos. A set of rotations are applied to all videos, and a pretext task is defined as prediction of these rotations. When accomplishing this task, 3DRotNet is actually trained to understand the semantic concepts and motions in videos. In other words, it learns a spatiotemporal video representation, which can be transferred to improve video understanding tasks in small datasets. Our extensive experiments successfully demonstrate the effectiveness of the proposed framework on action recognition, leading to significant improvements over the state-of-the-art self-supervised methods. With the self-supervised pre-trained 3DRotNet from large datasets, the recognition accuracy is boosted up by 20.4% on UCF101 and 16.7% on HMDB51 respectively, compared to the models trained from scratch.

Yang Xian, Yingli Tian

In this paper, a self-guiding multimodal LSTM (sg-LSTM) image captioning model is proposed to handle uncontrolled imbalanced real-world image-sentence dataset. We collect FlickrNYC dataset from Flickr as our testbed with 306,165 images and the original text descriptions uploaded by the users are utilized as the ground truth for training. Descriptions in FlickrNYC dataset vary dramatically ranging from short term-descriptions to long paragraph-descriptions and can describe any visual aspects, or even refer to objects that are not depicted. To deal with the imbalanced and noisy situation and to fully explore the dataset itself, we propose a novel guiding textual feature extracted utilizing a multimodal LSTM (m-LSTM) model. Training of m-LSTM is based on the portion of data in which the image content and the corresponding descriptions are strongly bonded. Afterwards, during the training of sg-LSTM on the rest training data, this guiding information serves as additional input to the network along with the image representations and the ground-truth descriptions. By integrating these input components into a multimodal block, we aim to form a training scheme with the textual information tightly coupled with the image content. The experimental results demonstrate that the proposed sg-LSTM model outperforms the traditional state-of-the-art multimodal RNN captioning framework in successfully describing the key components of the input images.