Building a Multi-modal Spatiotemporal Expert for Zero-shot Action Recognition with CLIP

• URL: http://arxiv.org/abs/2412.09895v2
• Date: Sun, 09 Feb 2025 12:42:37 GMT
• Title: Building a Multi-modal Spatiotemporal Expert for Zero-shot Action Recognition with CLIP
• Authors: Yating Yu, Congqi Cao, Yueran Zhang, Qinyi Lv, Lingtong Min, Yanning Zhang,
• Abstract summary: We propose a novel CLI framework to comprehend multi-temporal dynamics.<n>For the vision side, we propose an efficient Dynamic Crossshot Attention.<n>For the semantic side, we conduct text augmentation by constructing an Action Knowledge Graph.
• Score: 34.88916568947695
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Zero-shot action recognition (ZSAR) requires collaborative multi-modal spatiotemporal understanding. However, finetuning CLIP directly for ZSAR yields suboptimal performance, given its inherent constraints in capturing essential temporal dynamics from both vision and text perspectives, especially when encountering novel actions with fine-grained spatiotemporal discrepancies. In this work, we propose Spatiotemporal Dynamic Duo (STDD), a novel CLIP-based framework to comprehend multi-modal spatiotemporal dynamics synergistically. For the vision side, we propose an efficient Space-time Cross Attention, which captures spatiotemporal dynamics flexibly with simple yet effective operations applied before and after spatial attention, without adding additional parameters or increasing computational complexity. For the semantic side, we conduct spatiotemporal text augmentation by comprehensively constructing an Action Semantic Knowledge Graph (ASKG) to derive nuanced text prompts. The ASKG elaborates on static and dynamic concepts and their interrelations, based on the idea of decomposing actions into spatial appearances and temporal motions. During the training phase, the frame-level video representations are meticulously aligned with prompt-level nuanced text representations, which are concurrently regulated by the video representations from the frozen CLIP to enhance generalizability. Extensive experiments validate the effectiveness of our approach, which consistently surpasses state-of-the-art approaches on popular video benchmarks (i.e., Kinetics-600, UCF101, and HMDB51) under challenging ZSAR settings.

Related papers

• STOP: Integrated Spatial-Temporal Dynamic Prompting for Video Understanding [48.12128042470839]
  We propose an integrated Spatial-TempOral dynamic Prompting (STOP) model. It consists of two complementary modules, the intra-frame spatial prompting and inter-frame temporal prompting. STOP consistently achieves superior performance against state-of-the-art methods.
  arXiv  Detail & Related papers  (2025-03-20T09:16:20Z)
• EvolvingGS: High-Fidelity Streamable Volumetric Video via Evolving 3D Gaussian Representation [14.402479944396665]
  We introduce EvolvingGS, a two-stage strategy that first deforms the Gaussian model to align with the target frame, and then refines it with minimal point addition/subtraction. Owing to the flexibility of the incrementally evolving representation, our method outperforms existing approaches in terms of both per-frame and temporal quality metrics. Our method significantly advances the state-of-the-art in dynamic scene reconstruction, particularly for extended sequences with complex human performances.
  arXiv  Detail & Related papers  (2025-03-07T06:01:07Z)
• ARN-LSTM: A Multi-Stream Fusion Model for Skeleton-based Action Recognition [5.86850933017833]
  ARN-LSTM architecture is designed to address the challenge of simultaneously capturing spatial motion and temporal dynamics in action sequences.<n>Our proposed model integrates joint, motion, and temporal information through a multi-stream fusion architecture.
  arXiv  Detail & Related papers  (2024-11-04T03:29:51Z)
• Low-Light Video Enhancement via Spatial-Temporal Consistent Illumination and Reflection Decomposition [68.6707284662443]
  Low-Light Video Enhancement (LLVE) seeks to restore dynamic and static scenes plagued by severe invisibility and noise. One critical aspect is formulating a consistency constraint specifically for temporal-spatial illumination and appearance enhanced versions. We present an innovative video Retinex-based decomposition strategy that operates without the need for explicit supervision.
  arXiv  Detail & Related papers  (2024-05-24T15:56:40Z)
• MA-FSAR: Multimodal Adaptation of CLIP for Few-Shot Action Recognition [41.78245303513613]
  We introduce MA-FSAR, a framework that employs the Fine-Tuning (PEFT) technique to enhance the CLIP visual encoder in terms of action-related temporal and semantic representations. In addition to these token-level designs, we propose a prototype-level text-guided construction module to further enrich the temporal and semantic characteristics of video prototypes.
  arXiv  Detail & Related papers  (2023-08-03T04:17:25Z)
• Implicit Temporal Modeling with Learnable Alignment for Video Recognition [95.82093301212964]
  We propose a novel Implicit Learnable Alignment (ILA) method, which minimizes the temporal modeling effort while achieving incredibly high performance. ILA achieves a top-1 accuracy of 88.7% on Kinetics-400 with much fewer FLOPs compared with Swin-L and ViViT-H.
  arXiv  Detail & Related papers  (2023-04-20T17:11:01Z)
• Leaping Into Memories: Space-Time Deep Feature Synthesis [93.10032043225362]
  We propose LEAPS, an architecture-independent method for synthesizing videos from internal models. We quantitatively and qualitatively evaluate the applicability of LEAPS by inverting a range of architectures convolutional attention-based on Kinetics-400.
  arXiv  Detail & Related papers  (2023-03-17T12:55:22Z)
• Spatiotemporal Decouple-and-Squeeze Contrastive Learning for Semi-Supervised Skeleton-based Action Recognition [12.601122522537459]
  We propose a novel Stemporal Decouple Contrastive Learning (SDS-CL) framework to learn more abundant representations of skeleton-based actions. We present a new Temporal-squeezing Loss (STL), a new Temporal-squeezing Loss (TSL), and the Global-contrasting Loss (GL) to contrast the spatial-squeezing joint and motion features at the frame level, temporal-squeezing joint and motion features at the joint level, as well as global joint and motion features at the skeleton level.
  arXiv  Detail & Related papers  (2023-02-05T06:52:25Z)
• Learning Self-Similarity in Space and Time as Generalized Motion for Action Recognition [42.175450800733785]
  We propose a rich motion representation based on video self-similarity (STSS) We leverage the whole volume of STSSS and let our model learn to extract an effective motion representation from it. The proposed neural block, dubbed SELFY, can be easily inserted into neural architectures and trained end-to-end without additional supervision.
  arXiv  Detail & Related papers  (2021-02-14T07:32:55Z)
• Learn to cycle: Time-consistent feature discovery for action recognition [83.43682368129072]
  Generalizing over temporal variations is a prerequisite for effective action recognition in videos. We introduce Squeeze Re Temporal Gates (SRTG), an approach that favors temporal activations with potential variations. We show consistent improvement when using SRTPG blocks, with only a minimal increase in the number of GFLOs.
  arXiv  Detail & Related papers  (2020-06-15T09:36:28Z)
• A Spatial-Temporal Attentive Network with Spatial Continuity for Trajectory Prediction [74.00750936752418]
  We propose a novel model named spatial-temporal attentive network with spatial continuity (STAN-SC) First, spatial-temporal attention mechanism is presented to explore the most useful and important information. Second, we conduct a joint feature sequence based on the sequence and instant state information to make the generative trajectories keep spatial continuity.
  arXiv  Detail & Related papers  (2020-03-13T04:35:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.