Event perception tasks such as recognizing and localizing actions in streaming videos are essential for scaling to real-world application contexts. We tackle the problem of learning actor-centered representations through the notion of continual hierarchical predictive learning to localize actions in streaming videos without the need for training labels and outlines for the objects in the video. We propose a framework driven by the notion of hierarchical predictive learning to construct actor-centered features by attention-based contextualization. The key idea is that predictable features or objects do not attract attention and hence do not contribute to the action of interest. Experiments on three benchmark datasets show that the approach can learn robust representations for localizing actions using only one epoch of training, i.e., a single pass through the streaming video. We show that the proposed approach outperforms unsupervised and weakly supervised baselines while offering competitive performance to fully supervised approaches. Additionally, we extend the model to multi-actor settings to recognize group activities while localizing the multiple, plausible actors. We also show that it generalizes to out-of-domain data with limited performance degradation.
Multimodal self-supervised learning is getting more and more attention as it allows not only to train large networks without human supervision but also to search and retrieve data across various modalities. In this context, this paper proposes a self-supervised training framework that learns a common multimodal embedding space that, in addition to sharing representations across different modalities, enforces a grouping of semantically similar instances. To this end, we extend the concept of instance-level contrastive learning with a multimodal clustering step in the training pipeline to capture semantic similarities across modalities. The resulting embedding space enables retrieval of samples across all modalities, even from unseen datasets and different domains. To evaluate our approach, we train our model on the HowTo100M dataset and evaluate its zero-shot retrieval capabilities in two challenging domains, namely text-to-video retrieval, and temporal action localization, showing state-of-the-art results on four different datasets.
In temporal action localization methods, temporal downsampling operations are widely used to extract proposal features, but they often lead to the aliasing problem, due to lacking consideration of sampling rates. This paper aims to verify the existence of aliasing in TAL methods and investigate utilizing low pass filters to solve this problem by inhibiting the high-frequency band. However, the high-frequency band usually contains large amounts of specific information, which is important for model inference. Therefore, it is necessary to make a tradeoff between anti-aliasing and reserving high-frequency information. To acquire optimal performance, this paper learns different cutoff frequencies for different instances dynamically. This design can be plugged into most existing temporal modeling programs requiring only one additional cutoff frequency parameter. Integrating low pass filters to the downsampling operations significantly improves the detection performance and achieves comparable results on THUMOS'14, ActivityNet~1.3, and Charades datasets. Experiments demonstrate that anti-aliasing with low pass filters in TAL is advantageous and efficient.
Detection and localization of actions in videos is an important problem in practice. State-of-the-art video analytics systems are unable to efficiently and effectively answer such action queries because actions often involve a complex interaction between objects and are spread across a sequence of frames; detecting and localizing them requires computationally expensive deep neural networks. It is also important to consider the entire sequence of frames to answer the query effectively. In this paper, we present ZEUS, a video analytics system tailored for answering action queries. We present a novel technique for efficiently answering these queries using deep reinforcement learning. ZEUS trains a reinforcement learning agent that learns to adaptively modify the input video segments that are subsequently sent to an action classification network. The agent alters the input segments along three dimensions - sampling rate, segment length, and resolution. To meet the user-specified accuracy target, ZEUS's query optimizer trains the agent based on an accuracy-aware, aggregate reward function. Evaluation on three diverse video datasets shows that ZEUS outperforms state-of-the-art frame- and window-based filtering techniques by up to 22.1x and 4.7x, respectively. It also consistently meets the user-specified accuracy target across all queries.
This work strives for the classification and localization of human actions in videos, without the need for any labeled video training examples. Where existing work relies on transferring global attribute or object information from seen to unseen action videos, we seek to classify and spatio-temporally localize unseen actions in videos from image-based object information only. We propose three spatial object priors, which encode local person and object detectors along with their spatial relations. On top we introduce three semantic object priors, which extend semantic matching through word embeddings with three simple functions that tackle semantic ambiguity, object discrimination, and object naming. A video embedding combines the spatial and semantic object priors. It enables us to introduce a new video retrieval task that retrieves action tubes in video collections based on user-specified objects, spatial relations, and object size. Experimental evaluation on five action datasets shows the importance of spatial and semantic object priors for unseen actions. We find that persons and objects have preferred spatial relations that benefit unseen action localization, while using multiple languages and simple object filtering directly improves semantic matching, leading to state-of-the-art results for both unseen action classification and localization.
Weakly-supervised temporal action localization aims to localize action instances temporal boundary and identify the corresponding action category with only video-level labels. Traditional methods mainly focus on foreground and background frames separation with only a single attention branch and class activation sequence. However, we argue that apart from the distinctive foreground and background frames there are plenty of semantically ambiguous action context frames. It does not make sense to group those context frames to the same background class since they are semantically related to a specific action category. Consequently, it is challenging to suppress action context frames with only a single class activation sequence. To address this issue, in this paper, we propose an action-context modeling network termed ACM-Net, which integrates a three-branch attention module to measure the likelihood of each temporal point being action instance, context, or non-action background, simultaneously. Then based on the obtained three-branch attention values, we construct three-branch class activation sequences to represent the action instances, contexts, and non-action backgrounds, individually. To evaluate the effectiveness of our ACM-Net, we conduct extensive experiments on two benchmark datasets, THUMOS-14 and ActivityNet-1.3. The experiments show that our method can outperform current state-of-the-art methods, and even achieve comparable performance with fully-supervised methods. Code can be found at https://github.com/ispc-lab/ACM-Net
This paper introduces the task of few-shot common action localization in time and space. Given a few trimmed support videos containing the same but unknown action, we strive for spatio-temporal localization of that action in a long untrimmed query video. We do not require any class labels, interval bounds, or bounding boxes. To address this challenging task, we introduce a novel few-shot transformer architecture with a dedicated encoder-decoder structure optimized for joint commonality learning and localization prediction, without the need for proposals. Experiments on our reorganizations of the AVA and UCF101-24 datasets show the effectiveness of our approach for few-shot common action localization, even when the support videos are noisy. Although we are not specifically designed for common localization in time only, we also compare favorably against the few-shot and one-shot state-of-the-art in this setting. Lastly, we demonstrate that the few-shot transformer is easily extended to common action localization per pixel.
Weakly-supervised temporal action localization aims to localize actions in untrimmed videos with only video-level action category labels. Most of previous methods ignore the incompleteness issue of Class Activation Sequences (CAS), suffering from trivial localization results. To solve this issue, we introduce an adaptive mutual supervision framework (AMS) with two branches, where the base branch adopts CAS to localize the most discriminative action regions, while the supplementary branch localizes the less discriminative action regions through a novel adaptive sampler. The adaptive sampler dynamically updates the input of the supplementary branch with a sampling weight sequence negatively correlated with the CAS from the base branch, thereby prompting the supplementary branch to localize the action regions underestimated by the base branch. To promote mutual enhancement between these two branches, we construct mutual location supervision. Each branch leverages location pseudo-labels generated from the other branch as localization supervision. By alternately optimizing the two branches in multiple iterations, we progressively complete action regions. Extensive experiments on THUMOS14 and ActivityNet1.2 demonstrate that the proposed AMS method significantly outperforms the state-of-the-art methods.
We propose TubeR: a simple solution for spatio-temporal video action detection. Different from existing methods that depend on either an off-line actor detector or hand-designed actor-positional hypotheses like proposals or anchors, we propose to directly detect an action tubelet in a video by simultaneously performing action localization and recognition from a single representation. TubeR learns a set of tubelet-queries and utilizes a tubelet-attention module to model the dynamic spatio-temporal nature of a video clip, which effectively reinforces the model capacity compared to using actor-positional hypotheses in the spatio-temporal space. For videos containing transitional states or scene changes, we propose a context aware classification head to utilize short-term and long-term context to strengthen action classification, and an action switch regression head for detecting the precise temporal action extent. TubeR directly produces action tubelets with variable lengths and even maintains good results for long video clips. TubeR outperforms the previous state-of-the-art on commonly used action detection datasets AVA, UCF101-24 and JHMDB51-21.
Weakly-supervised Temporal Action Localization (WS-TAL) methods learn to localize temporal starts and ends of action instances in a video under only video-level supervision. Existing WS-TAL methods rely on deep features learned for action recognition. However, due to the mismatch between classification and localization, these features cannot distinguish the frequently co-occurring contextual background, i.e., the context, and the actual action instances. We term this challenge action-context confusion, and it will adversely affect the action localization accuracy. To address this challenge, we introduce a framework that learns two feature subspaces respectively for actions and their context. By explicitly accounting for action visual elements, the action instances can be localized more precisely without the distraction from the context. To facilitate the learning of these two feature subspaces with only video-level categorical labels, we leverage the predictions from both spatial and temporal streams for snippets grouping. In addition, an unsupervised learning task is introduced to make the proposed module focus on mining temporal information. The proposed approach outperforms state-of-the-art WS-TAL methods on three benchmarks, i.e., THUMOS14, ActivityNet v1.2 and v1.3 datasets.
Weakly-supervised temporal action localization (WS-TAL) aims to localize actions in untrimmed videos with only video-level labels. Most existing models follow the "localization by classification" procedure: locate temporal regions contributing most to the video-level classification. Generally, they process each snippet (or frame) individually and thus overlook the fruitful temporal context relation. Here arises the single snippet cheating issue: "hard" snippets are too vague to be classified. In this paper, we argue that learning by comparing helps identify these hard snippets and we propose to utilize snippet Contrastive learning to Localize Actions, CoLA for short. Specifically, we propose a Snippet Contrast (SniCo) Loss to refine the hard snippet representation in feature space, which guides the network to perceive precise temporal boundaries and avoid the temporal interval interruption. Besides, since it is infeasible to access frame-level annotations, we introduce a Hard Snippet Mining algorithm to locate the potential hard snippets. Substantial analyses verify that this mining strategy efficaciously captures the hard snippets and SniCo Loss leads to more informative feature representation. Extensive experiments show that CoLA achieves state-of-the-art results on THUMOS'14 and ActivityNet v1.2 datasets. CoLA code is publicly available at https://github.com/zhang-can/CoLA.
The object of Weakly-supervised Temporal Action Localization (WS-TAL) is to localize all action instances in an untrimmed video with only video-level supervision. Due to the lack of frame-level annotations during training, current WS-TAL methods rely on attention mechanisms to localize the foreground snippets or frames that contribute to the video-level classification task. This strategy frequently confuse context with the actual action, in the localization result. Separating action and context is a core problem for precise WS-TAL, but it is very challenging and has been largely ignored in the literature. In this paper, we introduce an Action-Context Separation Network (ACSNet) that explicitly takes into account context for accurate action localization. It consists of two branches (i.e., the Foreground-Background branch and the Action-Context branch). The Foreground- Background branch first distinguishes foreground from background within the entire video while the Action-Context branch further separates the foreground as action and context. We associate video snippets with two latent components (i.e., a positive component and a negative component), and their different combinations can effectively characterize foreground, action and context. Furthermore, we introduce extended labels with auxiliary context categories to facilitate the learning of action-context separation. Experiments on THUMOS14 and ActivityNet v1.2/v1.3 datasets demonstrate the ACSNet outperforms existing state-of-the-art WS-TAL methods by a large margin.
Temporal action localization (TAL) is a fundamental yet challenging task in video understanding. Existing TAL methods rely on pre-training a video encoder through action classification supervision. This results in a task discrepancy problem for the video encoder -- trained for action classification, but used for TAL. Intuitively, end-to-end model optimization is a good solution. However, this is not operable for TAL subject to the GPU memory constraints, due to the prohibitive computational cost in processing long untrimmed videos. In this paper, we resolve this challenge by introducing a novel low-fidelity end-to-end (LoFi) video encoder pre-training method. Instead of always using the full training configurations for TAL learning, we propose to reduce the mini-batch composition in terms of temporal, spatial or spatio-temporal resolution so that end-to-end optimization for the video encoder becomes operable under the memory conditions of a mid-range hardware budget. Crucially, this enables the gradient to flow backward through the video encoder from a TAL loss supervision, favourably solving the task discrepancy problem and providing more effective feature representations. Extensive experiments show that the proposed LoFi pre-training approach can significantly enhance the performance of existing TAL methods. Encouragingly, even with a lightweight ResNet18 based video encoder in a single RGB stream, our method surpasses two-stream ResNet50 based alternatives with expensive optical flow, often by a good margin.
Weakly-supervised Temporal Action Localization (WTAL) aims to detect the action segments with only video-level action labels in training. The key challenge is how to distinguish the action of interest segments from the background, which is unlabelled even on the video-level. While previous works treat the background as "curses", we consider it as "blessings". Specifically, we first use causal analysis to point out that the common localization errors are due to the unobserved confounder that resides ubiquitously in visual recognition. Then, we propose a Temporal Smoothing PCA-based (TS-PCA) deconfounder, which exploits the unlabelled background to model an observed substitute for the unobserved confounder, to remove the confounding effect. Note that the proposed deconfounder is model-agnostic and non-intrusive, and hence can be applied in any WTAL method without model re-designs. Through extensive experiments on four state-of-the-art WTAL methods, we show that the deconfounder can improve all of them on the public datasets: THUMOS-14 and ActivityNet-1.3.
Temporal action proposal generation aims to estimate temporal intervals of actions in untrimmed videos, which is a challenging yet important task in the video understanding field. The proposals generated by current methods still suffer from inaccurate temporal boundaries and inferior confidence used for retrieval owing to the lack of efficient temporal modeling and effective boundary context utilization. In this paper, we propose Temporal Context Aggregation Network (TCANet) to generate high-quality action proposals through "local and global" temporal context aggregation and complementary as well as progressive boundary refinement. Specifically, we first design a Local-Global Temporal Encoder (LGTE), which adopts the channel grouping strategy to efficiently encode both "local and global" temporal inter-dependencies. Furthermore, both the boundary and internal context of proposals are adopted for frame-level and segment-level boundary regressions, respectively. Temporal Boundary Regressor (TBR) is designed to combine these two regression granularities in an end-to-end fashion, which achieves the precise boundaries and reliable confidence of proposals through progressive refinement. Extensive experiments are conducted on three challenging datasets: HACS, ActivityNet-v1.3, and THUMOS-14, where TCANet can generate proposals with high precision and recall. By combining with the existing action classifier, TCANet can obtain remarkable temporal action detection performance compared with other methods. Not surprisingly, the proposed TCANet won the 1$^{st}$ place in the CVPR 2020 - HACS challenge leaderboard on temporal action localization task.
Temporal action localization is an important yet challenging task in video understanding. Typically, such a task aims at inferring both the action category and localization of the start and end frame for each action instance in a long, untrimmed video.While most current models achieve good results by using pre-defined anchors and numerous actionness, such methods could be bothered with both large number of outputs and heavy tuning of locations and sizes corresponding to different anchors. Instead, anchor-free methods is lighter, getting rid of redundant hyper-parameters, but gains few attention. In this paper, we propose the first purely anchor-free temporal localization method, which is both efficient and effective. Our model includes (i) an end-to-end trainable basic predictor, (ii) a saliency-based refinement module to gather more valuable boundary features for each proposal with a novel boundary pooling, and (iii) several consistency constraints to make sure our model can find the accurate boundary given arbitrary proposals. Extensive experiments show that our method beats all anchor-based and actionness-guided methods with a remarkable margin on THUMOS14, achieving state-of-the-art results, and comparable ones on ActivityNet v1.3. Code is available at https://github.com/TencentYoutuResearch/ActionDetection-AFSD.
Temporal action localization is an important and challenging task that aims to locate temporal regions in real-world untrimmed videos where actions occur and recognize their classes. It is widely acknowledged that video context is a critical cue for video understanding, and exploiting the context has become an important strategy to boost localization performance. However, previous state-of-the-art methods focus more on exploring semantic context which captures the feature similarity among frames or proposals, and neglect positional context which is vital for temporal localization. In this paper, we propose a temporal-position-sensitive context modeling approach to incorporate both positional and semantic information for more precise action localization. Specifically, we first augment feature representations with directed temporal positional encoding, and then conduct attention-based information propagation, in both frame-level and proposal-level. Consequently, the generated feature representations are significantly empowered with the discriminative capability of encoding the position-aware context information, and thus benefit boundary detection and proposal evaluation. We achieve state-of-the-art performance on both two challenging datasets, THUMOS-14 and ActivityNet-1.3, demonstrating the effectiveness and generalization ability of our method.
Real-world videos contain many complex actions with inherent relationships between action classes. In this work, we propose an attention-based architecture that models these action relationships for the task of temporal action localization in untrimmed videos. As opposed to previous works that leverage video-level co-occurrence of actions, we distinguish the relationships between actions that occur at the same time-step and actions that occur at different time-steps (i.e. those which precede or follow each other). We define these distinct relationships as action dependencies. We propose to improve action localization performance by modeling these action dependencies in a novel attention-based Multi-Label Action Dependency (MLAD)layer. The MLAD layer consists of two branches: a Co-occurrence Dependency Branch and a Temporal Dependency Branch to model co-occurrence action dependencies and temporal action dependencies, respectively. We observe that existing metrics used for multi-label classification do not explicitly measure how well action dependencies are modeled, therefore, we propose novel metrics that consider both co-occurrence and temporal dependencies between action classes. Through empirical evaluation and extensive analysis, we show improved performance over state-of-the-art methods on multi-label action localization benchmarks(MultiTHUMOS and Charades) in terms of f-mAP and our proposed metric.
This paper takes initial strides at designing and evaluating a vision-based system for privacy ensured activity monitoring. The proposed technology utilizing Artificial Intelligence (AI)-empowered proactive systems offering continuous monitoring, behavioral analysis, and modeling of human activities. To this end, this paper presents Single Run Action Detector (S-RAD) which is a real-time privacy-preserving action detector that performs end-to-end action localization and classification. It is based on Faster-RCNN combined with temporal shift modeling and segment based sampling to capture the human actions. Results on UCF-Sports and UR Fall dataset present comparable accuracy to State-of-the-Art approaches with significantly lower model size and computation demand and the ability for real-time execution on edge embedded device (e.g. Nvidia Jetson Xavier).
We introduce Activity Graph Transformer, an end-to-end learnable model for temporal action localization, that receives a video as input and directly predicts a set of action instances that appear in the video. Detecting and localizing action instances in untrimmed videos requires reasoning over multiple action instances in a video. The dominant paradigms in the literature process videos temporally to either propose action regions or directly produce frame-level detections. However, sequential processing of videos is problematic when the action instances have non-sequential dependencies and/or non-linear temporal ordering, such as overlapping action instances or re-occurrence of action instances over the course of the video. In this work, we capture this non-linear temporal structure by reasoning over the videos as non-sequential entities in the form of graphs. We evaluate our model on challenging datasets: THUMOS14, Charades, and EPIC-Kitchens-100. Our results show that our proposed model outperforms the state-of-the-art by a considerable margin.