Text-to-video diffusion models have made remarkable advancements. Driven by their ability to generate temporally coherent videos, research on zero-shot video editing using these fundamental models has expanded rapidly. To enhance editing quality, structural controls are frequently employed in video editing. Among these techniques, cross-attention mask control stands out for its effectiveness and efficiency. However, when cross-attention masks are naively applied to video editing, they can introduce artifacts such as blurring and flickering. Our experiments uncover a critical factor overlooked in previous video editing research: cross-attention masks are not consistently clear but vary with model structure and denoising timestep. To address this issue, we propose the metric Mask Matching Cost (MMC) that quantifies this variability and propose FreeMask, a method for selecting optimal masks tailored to specific video editing tasks. Using MMC-selected masks, we further improve the masked fusion mechanism within comprehensive attention features, e.g., temp, cross, and self-attention modules. Our approach can be seamlessly integrated into existing zero-shot video editing frameworks with better performance, requiring no control assistance or parameter fine-tuning but enabling adaptive decoupling of unedited semantic layouts with mask precision control. Extensive experiments demonstrate that FreeMask achieves superior semantic fidelity, temporal consistency, and editing quality compared to state-of-the-art methods.

As a fundamental backbone for video generation, diffusion models are challenged by low inference speed due to the sequential nature of denoising. Previous methods speed up the models by caching and reusing model outputs at uniformly selected timesteps. However, such a strategy neglects the fact that differences among model outputs are not uniform across timesteps, which hinders selecting the appropriate model outputs to cache, leading to a poor balance between inference efficiency and visual quality. In this study, we introduce Timestep Embedding Aware Cache (TeaCache), a training-free caching approach that estimates and leverages the fluctuating differences among model outputs across timesteps. Rather than directly using the time-consuming model outputs, TeaCache focuses on model inputs, which have a strong correlation with the modeloutputs while incurring negligible computational cost. TeaCache first modulates the noisy inputs using the timestep embeddings to ensure their differences better approximating those of model outputs. TeaCache then introduces a rescaling strategy to refine the estimated differences and utilizes them to indicate output caching. Experiments show that TeaCache achieves up to 4.41× acceleration over Open-Sora-Plan with negligible (-0.07% Vbench score) degradation of visual quality.

The current text-to-video (T2V) generation has made significant progress in synthesizing realistic general videos, but it is still under-explored in identity-specific human video generation with customized ID images. The key challenge lies in maintaining high ID fidelity consistently while preserving the original motion dynamic and semantic following after the identity injection. Current video identity customization methods mainly rely on reconstructing given identity images on text-to-image models, which have a divergent distribution with the T2V model. This process introduces a tuning-inference gap, leading to dynamic and semantic degradation. To tackle this problem, we propose a novel framework, dubbed $\textbf{PersonalVideo}$, that applies a mixture of reward supervision on synthesized videos instead of the simple reconstruction objective on images. Specifically, we first incorporate identity consistency reward to effectively inject the reference's identity without the tuning-inference gap. Then we propose a novel semantic consistency reward to align the semantic distribution of the generated videos with the original T2V model, which preserves its dynamic and semantic following capability during the identity injection. With the non-reconstructive reward training, we further employ simulated prompt augmentation to reduce overfitting by supervising generated results in more semantic scenarios, gaining good robustness even with only a single reference image. Extensive experiments demonstrate our method's superiority in delivering high identity faithfulness while preserving the inherent video generation qualities of the original T2V model, outshining prior methods.

Visual diffusion models achieve remarkable progress, yet they are typically trained at limited resolutions due to the lack of high-resolution data and constrained computation resources, hampering their ability to generate high-fidelity images or videos at higher resolutions. Recent efforts have explored tuning-free strategies to exhibit the untapped potential higher-resolution visual generation of pre-trained models. However, these methods are still prone to producing low-quality visual content with repetitive patterns. The key obstacle lies in the inevitable increase in high-frequency information when the model generates visual content exceeding its training resolution, leading to undesirable repetitive patterns deriving from the accumulated errors. To tackle this challenge, we propose FreeScale, a tuning-free inference paradigm to enable higher-resolution visual generation via scale fusion. Specifically, FreeScale processes information from different receptive scales and then fuses it by extracting desired frequency components. Extensive experiments validate the superiority of our paradigm in extending the capabilities of higher-resolution visual generation for both image and video models. Notably, compared with previous best-performing methods, FreeScale unlocks the 8k-resolution text-to-image generation for the first time.

Character image animation, which generates high-quality videos from a reference image and target pose sequence, has seen significant progress in recent years. However, most existing methods only apply to human figures, which usually do not generalize well on anthropomorphic characters commonly used in industries like gaming and entertainment. Our in-depth analysis suggests to attribute this limitation to their insufficient modeling of motion, which is unable to comprehend the movement pattern of the driving video, thus imposing a pose sequence rigidly onto the target character. To this end, this paper proposes $\texttt{Animate-X}$, a universal animation framework based on LDM for various character types (collectively named $\texttt{X}$), including anthropomorphic characters. To enhance motion representation, we introduce the Pose Indicator, which captures comprehensive motion pattern from the driving video through both implicit and explicit manner. The former leverages CLIP visual features of a driving video to extract its gist of motion, like the overall movement pattern and temporal relations among motions, while the latter strengthens the generalization of LDM by simulating possible inputs in advance that may arise during inference. Moreover, we introduce a new Animated Anthropomorphic Benchmark ($\texttt{$A^2$Bench}$) to evaluate the performance of $\texttt{Animate-X}$ on universal and widely applicable animation images. Extensive experiments demonstrate the superiority and effectiveness of $\texttt{Animate-X}$ compared to state-of-the-art methods.

Scaling visual generation models is essential for real-world content creation, yet requires substantial training and computational expenses. Alternatively, test-time scaling has garnered growing attention due to resource efficiency and promising performance. In this work, we present the first general test-time scaling framework for visual auto-regressive (VAR) models, TTS-VAR, modeling the generation process as a path searching problem. Inspired by VAR's hierarchical coarse-to-fine multi-scale generation, our framework integrates two key components: (i) At coarse scales, we observe that generated tokens are hard for evaluation, possibly leading to erroneous acceptance of inferior samples or rejection of superior samples. Noticing that the coarse scales contain sufficient structural information, we propose clustering-based diversity search. It preserves structural variety through semantic feature clustering, enabling later selection on samples with higher potential. (ii) In fine scales, resampling-based potential selection prioritizes promising candidates using potential scores, which are defined as reward functions incorporating multi-scale generation history. To dynamically balance computational efficiency with exploration capacity, we further introduce an adaptive descending batch size schedule throughout the causal generation process. Experiments on the powerful VAR model Infinity2B show a notable 8.7% GenEval score improvement (0.69→0.75). Key insights reveal that early-stage structural features effectively influence final quality, and resampling efficacy varies across generation scales.

We present Wan-Move, a simple and scalable framework that brings motion control to video generative models. Existing motion-controllable methods typically suffer from coarse control granularity and limited scalability, leaving their outputs insufficient for practical use. We narrow this gap by achieving precise and high-quality motion control. Our core idea is to directly make the original condition features motion-aware for guiding video synthesis. To this end, we first represent object motions with dense point trajectories, allowing fine-grained control over the scene. We then project these trajectories into latent space and propagate the first frame's features along each trajectory, producing an aligned spatiotemporal feature map that tells how each scene element should move. This feature map serves as the updated latent condition, which is naturally integrated into the off-the-shelf image-to-video model, e.g., Wan-I2V-14B, as motion guidance without any architecture change. It removes the need for auxiliary motion encoders and makes fine-tuning base models easily scalable. Through scaled training, Wan-Move generates 5-second, 480p videos whose motion controllability rivals Kling 1.5 Pro's commercial Motion Brush, as indicated by user studies. To support comprehensive evaluation, we further design MoveBench, a rigorously curated benchmark featuring diverse content categories and hybrid-verified annotations. It is distinguished by larger data volume, longer video durations, and high-quality motion annotations. Extensive experiments on MoveBench and the public dataset consistently show Wan-Move's superior motion quality. Code, models, and benchmark data are made available.

Customized generation using diffusion models has made impressive progress in image generation, but remains unsatisfactory in the challenging video generation task, as it requires the controllability of both subjects and motions. To that end, we present DreamVideo, a novel approach to generating personalized videos from a few static images of the desired subject and a few videos of target motion. DreamVideo decouples this task into two stages, subject learning and motion learning, by leveraging a pre-trained video diffusion model. The subject learning aims to accurately capture the fine appearance of the subject from provided images, which is achieved by combining textual inversion and fine-tuning of our carefully designed identity adapter. In motion learning, we architect a motion adapter and fine-tune it on the given videos to effectively model the target motion pattern. Combining these two lightweight and efficient adapters allows for flexible customization of any subject with any motion. Extensive experimental results demonstrate the superior performance of our DreamVideo over the state-of-the-art methods for customized video generation. Our project page is at https://dreamvideo-t2v.github.io.

Diffusion models have recently achieved remarkable progress in generating realistic images. However, chal-lenges remain in accurately understanding and synthesizing the layout requirements in the textual prompts. To align the generated image with layout instructions, we present a training-free layout calibration system S imM that inter-venes in the generative process on the fly during infer-ence time. Specifically, following a “check-locate-rectify” pipeline, the system first analyses the prompt to generate the target layout and compares it with the intermediate outputs to automatically detect errors. Then, by moving the located activations and making intra- and inter-map adjustments, the rectification process can be performed with negligible computational overhead. To evaluate SimM over a range of layout requirements, we present a benchmark SimMBench that compensates for the lack of superlative spatial relations in existing datasets. And both quantitative and qualitative results demonstrate the effectiveness of the proposed SimM in calibrating the layout inconsistencies. Our project page is at https://simm-t2i.github.io/SimM.

Despite diffusion models having shown powerful abilities to generate photorealistic images, generating videos that are realistic and diverse still remains in its infancy. One of the key reasons is that current methods intertwine spatial content and temporal dynamics together, leading to a notably increased complexity of text-to-video generation (T2V). In this work, we propose HiGen, a diffusion model-based method that improves performance by decoupling the spatial and temporal factors of videos from two perspectives, i.e., structure level and content level. At the structure level, we decompose the T2V task into two steps, including spatial reasoning and temporal reasoning, using a unified denoiser. Specifically, we generate spatially coherent priors using text during spatial reasoning and then generate temporally coherent motions from these priors during temporal reasoning. At the content level, we extract two subtle cues from the content of the input video that can express motion and appearance changes, respectively. These two cues then guide the model's training for generating videos, enabling flexible content variations and enhancing temporal stability. Through the decoupled paradigm, HiGen can effectively reduce the complexity of this task and generate realistic videos with semantics accuracy and motion stability. Extensive experiments demonstrate the superior performance of HiGen over the state-of-the-art T2V methods. We have released our source code and models.

Diffusion-based text-to-video generation has witnessed impressive progress in the past year yet still falls behind text-to-image generation. One of the key reasons is the limited scale of publicly available data (e.g., 10M video-text pairs in WebVid10m vs. 5B image-text pairs in LAION), considering the high cost of video captioning. Instead, it could be far easier to collect unlabeled clips from video platforms like YouTube. Motivated by this, we come up with a novel text-to-video generation framework, termed TF-T2V, which can directly learn with text-free videos. The rationale behind is to separate the process of text decoding from that of temporal modeling. To this end, we employ a content branch and a motion branch, which are jointly optimized with weights shared. Following such a pipeline, we study the effect of doubling the scale of training set (i.e., video-only WebVid10M) with some randomly collected text-free videos and are encouraged to observe the performance improvement (FID from 9.67 to 8.19 and FVD from 484 to 441), demonstrating the scalability of our approach. We also find that our model could enjoy sustainable performance gain (FID from 8.19 to 7.64 and FVD from 441 to 366) after reintroducing some text labels for training. Finally, we validate the effectiveness and generalizability of our ideology on both native text-to-video generation and compositional video synthesis paradigms. Code and models will be publicly available at here.

Diffusion models have emerged as the de facto paradigm for video generation. However, their reliance on web-scale data of varied quality often yields results that are visually unappealing and misaligned with the textual prompts. To tackle this problem, we propose InstructVideo to instruct text-to-video diffusion models with human feedback by reward fine-tuning. InstructVideo has two key ingredients: 1) To ameliorate the cost of reward fine-tuning induced by generating through the full DDIM sampling chain, we recast reward fine-tuning as editing. By leveraging the diffusion process to corrupt a sampled video, InstructVideo requires only partial inference of the DDIM sampling chain, reducing fine-tuning cost while improving fine-tuning efficiency. 2) To mitigate the absence of a dedicated video reward model for human preferences, we repurpose established image reward models, e.g., HPSv2. To this end, we propose Segmental Video Reward, a mechanism to provide reward signals based on segmental sparse sampling, and Temporally Attenuated Reward, a method that mitigates temporal modeling degradation during fine-tuning. Extensive experiments, both qualitative and quantitative, validate the practicality and efficacy of using image reward models in InstructVideo, significantly enhancing the visual quality of generated videos without compromising generalization capabilities. Code and models can be accessed through our project page https://instructvideo.github.io/.

Recent advancements in generation models have showcased remarkable capabilities in generating fantastic content. However, most of them are trained on proprietary high-quality data, and some models withhold their parameters and only provide accessible application programming interfaces (APIs), limiting their benefits for downstream tasks. To explore the feasibility of training a text-to-image generation model comparable to advanced models using publicly available resources, we introduce EvolveDirector. This framework interacts with advanced models through their public APIs to obtain text-image data pairs to train a base model. Our experiments with extensive data indicate that the model trained on generated data of the advanced model can approximate its generation capability. However, it requires large-scale samples of 10 million or more. This incurs significant expenses in time, computational resources, and especially the costs associated with calling fee-based APIs. To address this problem, we leverage pre-trained large vision-language models (VLMs) to guide the evolution of the base model. VLM continuously evaluates the base model during training and dynamically updates and refines the training dataset by the discrimination, expansion, deletion, and mutation operations. Experimental results show that this paradigm significantly reduces the required data volume. Furthermore, when approaching multiple advanced models, EvolveDirector can select the best samples generated by them to learn powerful and balanced abilities. The final trained model Edgen is demonstrated to outperform these advanced models. The code and model weights are available at https://github.com/showlab/EvolveDirector.

Standard approaches for video action recognition usually operate on full input videos, which is inefficient due to the widespread spatio-temporal redundancy in videos. The recent progress in masked video modelling, specifically VideoMAE, has shown the ability of vanilla Vision Transformers (ViT) to complement spatio-temporal contexts using limited visual content. Inspired by this, we propose Masked Action Recognition (MAR), which reduces redundant computation by discarding a proportion of patches and operating only on a portion of the videos. MAR includes two essential components:cell running maskingandbridging classifier. Specifically, to enable the ViT to perceive the details beyond the visible patches, cell running masking is used to preserve the spatio-temporal correlations in videos. This ensures that the patches at the same spatial location can be observed in turn for easy reconstructions. Additionally, we notice that, although the partially observed features can reconstruct semantically explicit invisible patches, they fail to achieve accurate classification. To address this issue, we propose a bridging classifier that can help fill the semantic gap between the ViT encoded features used for reconstruction and the specialized features used for classification. Our proposed MAR can reduce the computational cost of ViT by 53%. Extensive experiments have demonstrated that MAR consistently outperforms existing ViT models by a notable margin. Notably, we found that a ViT-Large model fine-tuned by MAR achieves comparable performance to a ViT-Huge model fine-tuned by standard training methods on both Kinetics-400 and Something-Something v2 datasets. Moreover, the computation overhead of our ViT-Large model is only 14.5% of that of the ViT-Huge model. Codes have been made availablehttps://github.com/alibaba-mmai-research/Masked-Action-Recognition.

Open-set action recognition is to reject unknown human action cases which are out of the distribution of the training set. Existing methods mainly focus on learning better uncertainty scores but dismiss the importance of feature representations. We find that features with richer semantic diversity can significantly improve the open-set performance under the same uncertainty scores. In this paper, we begin with analyzing the feature representation behavior in the open-set action recognition (OSAR) problem based on the information bottleneck (IB) theory, and propose to enlarge the instance-specific (IS) and classspecific (CS) information contained in the feature for better performance. To this end, a novel Prototypical Similarity Learning (PSL) framework is proposed to keep the instance variance within the same class to retain more IS information. Besides, we notice that unknown samples sharing similar appearances to known samples are easily misclassified as known classes. To alleviate this issue, video shuffling is further introduced in our PSL to learn distinct temporal information between original and shuffled samples, which we find enlarges the CS information. Extensive experiments demonstrate that the proposed PSL can significantly boost both the open-set and closed-set performance and achieves state-of-the-art results on multiple benchmarks. Code is available at https://github.com/Jun-CEN/PSL.

Current state-of-the-art approaches for few-shot action recognition achieve promising performance by conducting frame-level matching on learned visual features. However, they generally suffer from two limitations: i) the matching procedure between local frames tends to be inaccurate due to the lack of guidance to force long-range temporal perception; ii) explicit motion learning is usually ignored, leading to partial information loss. To address these issues, we develop a Motion-augmented Long-short Contrastive Learning (MoLo) method that contains two crucial components, including a long-short contrastive objective and a motion autodecoder. Specifically, the long-short contrastive objective is to endow local frame features with long-form temporal awareness by maximizing their agreement with the global token of videos belonging to the same class. The motion autodecoder is a lightweight architecture to reconstruct pixel motions from the differential features, which explicitly embeds the network with motion dynamics. By this means, MoLo can simultaneously learn long-range temporal context and motion cues for comprehensive few-shot matching. To demonstrate the effectiveness, we evaluate MoLo on five standard benchmarks, and the results show that MoLo favorably outperforms recent advanced methods. The source code is available at https://github.com/alibaba-mmai-research/MoLo.

Generating a talking face video from the input audio sequence is a practical yet challenging task. Most existing methods either fail to capture fine facial details or need to train a specific model for each identity. We argue that a codebook pre-learned on high-quality face images can serve as a useful prior that facilitates high-fidelity and generalizable talking head synthesis. Thanks to the strong capability of the codebook in representing face textures, we simplify the talking face generation task as finding proper lip-codes to characterize the variation of lips during portrait talking. To this end, we propose LipFormer, a Transformer-based framework to model the audio-visual coherence and predict the lip-codes sequence based on input audio features. We further introduce an adaptive face warping module, which helps warp the reference face to the target pose in the feature space, to alleviate the difficulty of lip-code prediction under different poses. By this means, LipFormer can make better use of prelearned priors in images and is robust to posture change. Extensive experiments show that LipFormer can produce more realistic talking face videos compared to previous methods and faithfully generalize to unseen identities.

Recently, prompt-based learning has made impressive progress on image class-incremental learning, but it still lacks sufficient exploration in the video domain. In this paper, we will fill this gap by learning multiple prompts based on a powerful image-language pre-trained model, i.e., CLIP, making it fit for video class-incremental learning (VCIL). For this purpose, we present a space-time prompting approach (ST-Prompt) which contains two kinds of prompts, i.e., task-specific prompts and task-agnostic prompts. The task-specific prompts are to address the catastrophic forgetting problem by learning multi-grained prompts, i.e., spatial prompts, temporal prompts and comprehensive prompts, for accurate task identification. The task-agnostic prompts maintain a globally-shared prompt pool, which can empower the pre-trained image models with temporal perception abilities by exchanging contexts between frames. By this means, ST-Prompt can transfer the plentiful knowledge in the image-language pre-trained models to the VCIL task with only a tiny set of prompts to be optimized. To evaluate ST-Prompt, we conduct extensive experiments on three standard benchmarks. The results show that ST-Prompt can significantly surpass the state-of-the-art VCIL methods, especially it gains 9.06% on HMDB51 dataset under the 1 × 25 stage setting.

Recently, large-scale pre-trained language-image models like CLIP have shown extraordinary capabilities for understanding spatial contents, but naively transferring such models to video recognition still suffers from unsatisfactory temporal modeling capabilities. Existing methods insert tunable structures into or in parallel with the pre-trained model, which either requires back-propagation through the whole pre-trained model and is thus resource-demanding, or is limited by the temporal reasoning capability of the pre-trained structure. In this work, we present DiST, which disentangles the learning of spatial and temporal aspects of videos. Specifically, DiST uses a dual-encoder structure, where a pre-trained foundation model acts as the spatial encoder, and a lightweight network is introduced as the temporal encoder. An integration branch is inserted between the encoders to fuse spatio-temporal information. The disentangled spatial and temporal learning in DiST is highly efficient because it avoids the back-propagation of massive pre-trained parameters. Meanwhile, we empirically show that disentangled learning with an extra network for integration benefits both spatial and temporal understanding. Extensive experiments on five benchmarks show that DiST delivers better performance than existing state-of-the-art methods by convincing gaps. When pre-training on the large-scale Kinetics-710, we achieve 89.7% on Kinetics-400 with a frozen ViT-L model, which verifies the scalability of DiST. Codes and models can be found in https://github.com/alibaba-mmai-research/DiST.

Relational Language-Image Pre-training (RLIP) aims to align vision representations with relational texts, thereby advancing the capability of relational reasoning in computer vision tasks. However, hindered by the slow convergence of RLIPv11architecture and the limited availability of existing scene graph data, scaling RLIPv1 is challenging. In this paper, we propose RLIPv2, a fast converging model that enables the scaling of relational pre-training to large-scale pseudo-labelled scene graph data. To enable fast scaling, RLIPv2 introduces Asymmetric Language-Image Fusion (ALIF), a mechanism that facilitates earlier and deeper gated cross-modal fusion with sparsified language encoding layers. ALIF leads to comparable or better performance than RLIPv1 in a fraction of the time for pre-training and fine-tuning. To obtain scene graph data at scale, we extend object detection datasets with free-form relation labels by introducing a captioner (e.g., BLIP) and a designed Relation Tagger. The Relation Tagger assigns BLIP-generated relation texts to region pairs, thus enabling larger-scale relational pre-training. Through extensive experiments conducted on Human-Object Interaction Detection and Scene Graph Generation, RLIPv2 shows state-of-the-art performance on three benchmarks under fully-finetuning, few-shot and zero-shot settings. Notably, the largest RLIPv2 achieves 23.29mAP on HICO-DET without any fine-tuning, yields 32.22mAP with just 1% data and yields 45.09mAP with 100% data. Code and models are publicly available at https://github.com/JacobYuan7/RLIPv2.

The pursuit of controllability as a higher standard of visual content creation has yielded remarkable progress in customizable image synthesis. However, achieving controllable video synthesis remains challenging due to the large variation of temporal dynamics and the requirement of cross-frame temporal consistency. Based on the paradigm of compositional generation, this work presents VideoComposer that allows users to flexibly compose a video with textual conditions, spatial conditions, and more importantly temporal conditions. Specifically, considering the characteristic of video data, we introduce the motion vector from compressed videos as an explicit control signal to provide guidance regarding temporal dynamics. In addition, we develop a Spatio-Temporal Condition encoder (STC-encoder) that serves as a unified interface to effectively incorporate the spatial and temporal relations of sequential inputs, with which the model could make better use of temporal conditions and hence achieve higher inter-frame consistency. Extensive experimental results suggest that VideoComposer is able to control the spatial and temporal patterns simultaneously within a synthesized video in various forms, such as text description, sketch sequence, reference video, or even simply hand-crafted motions. The code and models are publicly available at https://videocomposer.github.io.

This work presents FaceComposer, a unified generative model that accomplishes a variety of facial content creation tasks, including text-conditioned face synthesis, text-guided face editing, face animation etc. Based on the latent diffusion framework, FaceComposer follows the paradigm of compositional generation and employs diverse face-specific conditions, e.g., Identity Feature and Projected Normalized Coordinate Code, to release the model creativity at all possible. To support text control and animation, we clean up some existing face image datasets and collect around 500 hours of talking-face videos, forming a high-quality large-scale multi-modal face database. A temporal self-attention module is incorporated into the U-Net structure, which allows learning the denoising process on the mixture of images and videos. Extensive experiments suggest that our approach not only achieves comparable or even better performance than state-of-the-arts on each single task, but also facilitates some combined tasks with one-time forward, demonstrating its potential in serving as a foundation generative model in face domain. We further develop an interface such that users can enjoy our one-step service to create, edit, and animate their own characters. Code, dataset, model, and interface will be made publicly available.

Visual tracking has made significant improvements in the past few decades. Most existing state-of-the-art trackers 1) merely aim for performance in ideal conditions while overlooking the real-world conditions; 2) adopt the tracking-by-detection paradigm, neglecting rich temporal contexts; 3) only integrate the temporal information into the template, where temporal contexts among consecutive frames are far from being fully utilized. To handle those problems, we propose a two-level framework (TCTrack) that can exploit temporal contexts efficiently. Based on it, we propose a stronger version for real-world visual tracking, i.e., TCTrack++. It boils down to two levels: features and similarity maps. Specifically, for feature extraction, we propose an attention-based temporally adaptive convolution to enhance the spatial features using temporal information, which is achieved by dynamically calibrating the convolution weights. For similarity map refinement, we introduce an adaptive temporal transformer to encode the temporal knowledge efficiently and decode it for the accurate refinement of the similarity map. To further improve the performance, we additionally introduce a curriculum learning strategy. Also, we adopt online evaluation to measure performance in real-world conditions. Exhaustive experiments on 8 well-known benchmarks demonstrate the superiority of TCTrack++. Real-world tests directly verify that TCTrack++ can be readily used in real-world applications.

Natural untrimmed videos provide rich visual content for self-supervised learning. Yet most previous efforts to learn spatio-temporal representations rely on manually trimmed videos, such as Kinetics dataset (Carreira and Zisserman 2017), resulting in limited diversity in visual patterns and limited performance gains. In this work, we aim to improve video representations by leveraging the rich information in natural untrimmed videos. For this purpose, we propose learning a hierarchy of temporal consistencies in videos, i.e., visual consistency and topical consistency, corresponding respectively to clip pairs that tend to be visually similar when separated by a short time span, and clip pairs that share similar topics when separated by a long time span. Specifically, we present a Hierarchical Consistency (HiCo++) learning framework, in which the visually consistent pairs are encouraged to share the same feature representations by contrastive learning, while topically consistent pairs are coupled through a topical classifier that distinguishes whether they are topic-related, i.e., from the same untrimmed video. Additionally, we impose a gradual sampling algorithm for the proposed hierarchical consistency learning, and demonstrate its theoretical superiority. Empirically, we show that HiCo++ can not only generate stronger representations on untrimmed videos, but also improve the representation quality when applied to trimmed videos. This contrasts with standard contrastive learning, which fails to learn powerful representations from untrimmed videos. Source code will be made available here.

The central idea of contrastive learning is to discriminate between different instances and force different views from the same instance to share the same representation. To avoid trivial solutions, augmentation plays an important role in generating different views, among which random cropping is shown to be effective for the model to learn a generalized and robust representation. Commonly used random crop operation keeps the distribution of the difference between two views unchanged along the training process. In this work, we show that adaptively controlling the disparity between two augmented views along the training process enhances the quality of the learned representations. Specifically, we present a parametric cubic cropping operation, ParamCrop, for video contrastive learning, which automatically crops a 3D cubic by differentiable 3D affine transformations. ParamCrop is trained simultaneously with the video backbone using an adversarial objective, so that it learns to increase the contrastive loss and thus gradually reduces the shared contents between two cropped views. Experiments show that this adaptive and gradual increase in the disparity yielded by ParamCrop is beneficial to learning a strong and generalized representation for downstream tasks, which is shown to be effective on multiple contrastive learning frameworks and video backbones.

Temporal contexts among consecutive frames are far from being fully utilized in existing visual trackers. In this work, we present TCTrack11https://github.com/vision4robotics/TCTrack, a comprehensive framework to fully exploit temporal contexts for aerial tracking. The temporal contexts are incorporated at two levels: the extraction of features and the refinement of similarity maps. Specifically, for feature extraction, an online temporally adaptive convolution is proposed to enhance the spatial features using temporal information, which is achieved by dynamically calibrating the convolution weights according to the previous frames. For similarity map refinement, we propose an adaptive temporal transformer, which first effectively encodes temporal knowledge in a memory-efficient way, before the temporal knowledge is decoded for accurate adjustment of the similarity map. TCTrack is effective and efficient: evaluation on four aerial tracking benchmarks shows its impressive performance; real-world UAV tests show its high speed of over 27 FPS on NVIDIA Jetson AGX Xavier.

Natural videos provide rich visual contents for selfsupervised learning. Yet most existing approaches for learning spatio-temporal representations rely on manually trimmed videos, leading to limited diversity in visual patterns and limited performance gain. In this work, we aim to learn representations by leveraging more abundant information in untrimmed videos. To this end, we propose to learn a hierarchy of consistencies in videos, i.e., visual consistency and topical consistency, corresponding respectively to clip pairs that tend to be visually similar when separated by a short time span and share similar topics when separated by a long time span. Specifically, a hierarchical consistency learning framework HiCo is presented, where the visually consistent pairs are encouraged to have the same representation through contrastive learning, while the topically consistent pairs are coupled through a topical classifier that distinguishes whether they are topicrelated. Further, we impose a gradual sampling algorithm for proposed hierarchical consistency learning, and demonstrate its theoretical superiority. Empirically, we show that not only HiCo can generate stronger representations on untrimmed videos, it also improves the representation quality when applied to trimmed videos. This is in contrast to standard contrastive learning that fails to learn appropriate representations from untrimmed videos.

Current few-shot action recognition methods reach impressive performance by learning discriminative features for each video via episodic training and designing various temporal alignment strategies. Nevertheless, they are limited in that (a) learning individual features without considering the entire task may lose the most relevant information in the current episode, and (b) these alignment strategies may fail in misaligned instances. To overcome the two limitations, we propose a novel Hybrid Relation guided Set Matching (HyRSM) approach that incorporates two key components: hybrid relation module and set matching metric. The purpose of the hybrid relation module is to learn task-specific embeddings by fully exploiting associated relations within and cross videos in an episode. Built upon the task-specific features, we reformulate distance measure between query and support videos as a set matching problem and further design a bidirectional Mean Hausdorff Metric to improve the resilience to misaligned instances. By this means, the proposed HyRSM can be highly informative and flexible to predict query categories under the few-shot settings. We evaluate HyRSM on six challenging benchmarks, and the experimental results show its superiority over the state-of-the-art methods by a convincing margin. Project page: https://hyrsm-cvpr2022.github.io/.

Recent incremental learning for action recognition usually stores representative videos to mitigate catastrophic forgetting. However, only a few bulky videos can be stored due to the limited memory. To address this problem, we propose FrameMaker, a memory-efficient video class-incremental learning approach that learns to produce a condensed frame for each selected video. Specifically, FrameMaker is mainly composed of two crucial components: Frame Condensing and Instance-Specific Prompt. The former is to reduce the memory cost by preserving only one condensed frame instead of the whole video, while the latter aims to compensate the lost spatio-temporal details in the Frame Condensing stage. By this means, FrameMaker enables a remarkable reduction in memory but keep enough information that can be applied to following incremental tasks. Experimental results on multiple challenging benchmarks, i.e., HMDB51, UCF101 and Something-Something V2, demonstrate that FrameMaker can achieve better performance to recent advanced methods while consuming only 20% memory. Additionally, under the same memory consumption conditions, FrameMaker significantly outperforms existing state-of-the-arts by a convincing margin.

The understanding of language plays a key role in video grounding, where a target moment is localized according to a text query. From a biological point of view, language is naturally hierarchical, with the main clause (predicate phrase) providing coarse semantics and modifiers providing detailed descriptions. In video grounding, moments described by the main clause may exist in multiple clips of a long video, including both the ground-truth and background clips. Therefore, in order to correctly discriminate the ground-truth clip from the background ones, this co-existence leads to the negligence of the main clause, and concentrate the model on the modifiers that provide discriminative information on distinguishing the target proposal from the others. We first demonstrate this phenomenon empirically, and propose a Hierarchical Language Network (HLN) that exploits the language hierarchy, as well as a new learning approach called Multi-Instance Positive-Unlabelled Learning (MI-PUL) to alleviate the above problem. Specifically, in HLN, the localization is performed on various layers of the language hierarchy, so that the attention can be paid to different parts of the sentences, rather than only discriminative ones. Furthermore, MI-PUL allows the model to localize background clips that can be possibly described by the main clause, even without manual annotations. Therefore, the union of the two proposed components enhances the learning of the main clause, which is of critical importance in video grounding. Finally, we evaluate that our proposed HLN can plug into the current methods and improve their performance. Extensive experiments on challenging datasets show HLN significantly improve the state-of-the-art methods, especially achieving 6.15% gain in terms of [Formula: see text] on the TACoS dataset.

Temporal action detection (TAD) aims to determine the semantic label and the temporal interval of every action instance in an untrimmed video. It is a fundamental and challenging task in video understanding. Previous methods tackle this task with complicated pipelines. They often need to train multiple networks and involve hand-designed operations, such as non-maximal suppression and anchor generation, which limit the flexibility and prevent end-to-end learning. In this paper, we propose an end-to-end Transformer-based method for TAD, termed TadTR. Given a small set of learnable embeddings called action queries, TadTR adaptively extracts temporal context information from the video for each query and directly predicts action instances with the context. To adapt Transformer to TAD, we propose three improvements to enhance its locality awareness. The core is a temporal deformable attention module that selectively attends to a sparse set of key snippets in a video. A segment refinement mechanism and an actionness regression head are designed to refine the boundaries and confidence of the predicted instances, respectively. With such a simple pipeline, TadTR requires lower computation cost than previous detectors, while preserving remarkable performance. As a self-contained detector, it achieves state-of-the-art performance on THUMOS14 (56.7% mAP) and HACS Segments (32.09% mAP). Combined with an extra action classifier, it obtains 36.75% mAP on ActivityNet-1.3. Code is available at https://github.com/xlliu7/TadTR.

Motions are reflected in videos as the movement of pixels, and actions are essentially patterns of inconsistent motions between the foreground and the background. To well distinguish the actions, especially those with complicated spatio-temporal interactions, correctly locating the prominent motion areas is of crucial importance. However, most motion information in existing videos are difficult to label and training a model with good motion representations with supervision will thus require a large amount of human labour for annotation. In this paper, we address this problem by self-supervised learning. Specifically, we propose to learn Motion from Static Images (MoSI). The model learns to encode motion information by classifying pseudo motions generated by MoSI. We furthermore introduce a static mask in pseudo motions to create local motion patterns, which forces the model to additionally locate notable motion areas for the correct classification. We demonstrate that MoSI can discover regions with large motion even without fine-tuning on the downstream datasets. As a result, the learned motion representations boost the performance of tasks requiring understanding of complex scenes and motions, i.e., action recognition. Extensive experiments show the consistent and transferable improvements achieved by MoSI. Codes will be soon released.

Self-supervised learning presents a remarkable performance to utilize unlabeled data for various video tasks. In this paper, we focus on applying the power of self-supervised methods to improve semi-supervised action proposal generation. Particularly, we design an effective Self-supervised Semi-supervised Temporal Action Proposal (SSTAP) framework. The SSTAP contains two crucial branches, i.e., temporal-aware semi-supervised branch and relation-aware self-supervised branch. The semi-supervised branch improves the proposal model by introducing two temporal perturbations, i.e., temporal feature shift and temporal feature flip, in the mean teacher framework. The self-supervised branch defines two pretext tasks, including masked feature reconstruction and clip-order prediction, to learn the relation of temporal clues. By this means, SSTAP can better explore unlabeled videos, and improve the discriminative abilities of learned action features. We extensively evaluate the proposed SSTAP on THUMOS14 and ActivityNet v1.3 datasets. The experimental results demonstrate that SSTAP significantly outperforms state-of-the-art semi-supervised methods and even matches fully-supervised methods. Code is available at https://github.com/wangxiang1230/SSTAP.

Current approaches for video grounding propose kinds of complex architectures to capture the video-text relations, and have achieved impressive improvements. However, it is hard to learn the complicated multi-modal relations by only architecture designing in fact. In this paper, we introduce a novel Support-set Based Cross-Supervision (Sscs) module which can improve existing methods during training phase without extra inference cost. The proposed Sscs module contains two main components, i.e., discriminative contrastive objective and generative caption objective. The contrastive objective aims to learn effective representations by contrastive learning, while the caption objective can train a powerful video encoder supervised by texts. Due to the co-existence of some visual entities in both ground-truth and background intervals, i.e. mutual exclusion, naively contrastive learning is unsuitable to video grounding. We address the problem by boosting the cross-supervision with the support-set concept, which collects visual information from the whole video and eliminates the mutual exclusion of entities. Combined with the original objectives, Sscs can enhance the abilities of multi-modal relation modeling for existing approaches. We extensively evaluate Sscs on three challenging datasets, and show that our method can improve current state-of-the-art methods by large margins, especially 6.35% in terms of [email protected] on Charades-STA.

Most recent approaches for online action detection tend to apply Recurrent Neural Network (RNN) to capture long-range temporal structure. However, RNN suffers from non-parallelism and gradient vanishing, hence it is hard to be optimized. In this paper, we propose a new encoder-decoder framework based on Transformers, named OadTR, to tackle these problems. The encoder attached with a task token aims to capture the relationships and global inter-actions between historical observations. The decoder extracts auxiliary information by aggregating anticipated future clip representations. Therefore, OadTR can recognize current actions by encoding historical information and predicting future context simultaneously. We extensively evaluate the proposed OadTR on three challenging datasets: HDD, TVSeries, and THUMOS14. The experimental results show that OadTR achieves higher training and inference speeds than current RNN based approaches, and significantly outperforms the state-of-the-art methods in terms of both mAP and mcAP. Code is available at https://github.com/wangxiang1230/OadTR.

In this paper, we address the challenging problem of weakly supervised spatio-temporal action localization for which only video-level action labels are available during training. To solve this problem, we propose an end-to-end Global Local Network (GLNet) to predict the probability distribution simultaneously in both spatial and temporal space. The proposed GLNet model includes two key components: a local spatial module and a global temporal module. The local spatial module aims to predict the frame-level spatial distribution by encoding short-term temporal information. In particular, we propose a Region Actionness Network (RAN) to select the target region boxes from the precomputed exhaustive proposals. The global temporal module can predict temporal distribution by a long-term temporal structure modelling. Specifically, we design a temporal fusion-and-excitation architecture on the top of several clips, and trained by a sparse loss function. Therefore, the proposed GLNet model can perform spatio-temporal action localization in an end-to-end manner. We evaluate the performance of GLNet on the J-HMDB and UCF101-24 datasets. The experimental results demonstrate GLNet achieves a significant margin against other state-of-the-art weakly supervised methods and even some fully supervised methods in terms of frame mean Average Precision (mAP) and the video mAP (called frame-mAP and video-mAP, respectively).

Current state-of-the-art approaches for spatio-temporal action detection have achieved impressive results but remain unsatisfactory for temporal extent detection. The main reason comes from that, there are some ambiguous states similar to the real actions which may be treated as target actions even by a well trained network. In this paper, we define these ambiguous samples as “transitional states”, and propose a Transition-Aware Context Network (TACNet) to distinguish transitional states. The proposed TACNet includes two main components, i.e., temporal context detector and transition-aware classifier. The temporal context detector can extract long-term context information with constant time complexity by constructing a recurrent network. The transition-aware classifier can further distinguish transitional states by classifying action and transitional states simultaneously. Therefore, the proposed TACNet can substantially improve the performance of spatio-temporal action detection. We extensively evaluate the proposed TACNet on UCF101-24 and J-HMDB datasets. The experimental results demonstrate that TACNet obtains competitive performance on JHMDB and significantly outperforms the state-of-the-art methods on the untrimmed UCF101 24 in terms of both frame-mAP and video-mAP.

Semantic parts have shown a powerful discriminative capacity for action recognition. However, many existing methods select parts according to predefined heuristic rules, which may cause the correlation among parts to be lost, or do not appropriately consider the cluttered candidate part space, which may result in weak generalizability of the resulting action labels. Therefore, better consideration of the correlation among parts and refinement of the candidate space will lead to a more discriminative action representation. This paper achieves improved performance by more elegantly addressing these two factors. First, considering the cluttered nature of the candidate space, we propose a recursive part elimination strategy for iterative refinement of the candidate parts. In each iteration, we eliminate the parts with the lowest weights, which are deemed to be noise. Second, we measure the discriminative capabilities of the candidates and select the top-ranked parts by applying a maximum margin model, which can alleviate overfitting while simultaneously improving generalizability and correlation extraction. Finally, using the selected parts, we extract mid-level features. We report experiments conducted on four datasets (KTH, Olympic Sports, UCF50, and HMDB51). The proposed method can achieve significant improvements compared with other recent methods, including a lower computational cost, a faster speed, and higher accuracy.

Mid-level parts are shown to be effective for human action recognition in videos. Typically, these semantic parts are first mined with some heuristic rules, then videos are represented via volumetric max-pooling (VMP) method. However, these methods have two issues: 1) the VMP strategy divides videos by static grids. In this case, a semantic part may occur in different localizations in different videos. That means the VMP strategy loses the space-time invariance. To solve this problem, we propose to apply a saliency-driven max-pooling scheme to represent a video. We extract the video semantic cues by the saliency map, and dynamically pool the local maximum responses. This scheme can be considered as a semantic content-based feature alignment method and 2) the parts discovered by heuristic rules may be intuitive but not discriminative enough for action classification because they neglect the relations between the detectors. For this issue, we propose to apply a sparse classifier model to select discriminative parts. Moreover, to further improve the discriminative ability of the representation, we propose to conduct feature selection by the corresponding entry magnitude of the model coefficients. We conduct experiments on four challenging datasets-KTH, Olympic Sports, UCF50, and HMDB51. The results show that the proposed method significantly outperforms the state-of-the-art methods.