arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2304.08451cs.CV
收藏

基于Token丢弃与上下文精炼的高效视频动作检测

Efficient Video Action Detection with Token Dropout and Context Refinement

Nanjing University(南京大学);Ant Group(蚂蚁集团);Fudan University(复旦大学);Shanghai AI Lab(上海人工智能实验室)

Lei Chen, Zhan Tong, Yibing Song, Gangshan Wu, Limin Wang

AI总结:

针对视频动作检测中ViT因token规模大导致效率低的问题,提出含时空token丢弃和上下文精炼的EVAD框架,在保证精度的同时显著降低计算量、提升推理速度。

摘要:

大规模视频token的流式视频片段阻碍了视觉Transformer(ViTs)实现高效识别,尤其是在需要充足时空表征来精准识别动作执行者的视频动作检测任务中。本研究提出了一个基于原生ViTs的端到端高效视频动作检测框架EVAD。我们的EVAD包含两个针对视频动作检测的专门设计。首先,我们从以关键帧为中心的视角提出了时空token丢弃策略:在一个视频片段中,我们保留关键帧的所有token,保留其他帧中与动作执行者运动相关的token,丢弃该片段中其余的token。其次,我们利用剩余token精炼场景上下文,以更好地识别动作执行者身份。我们将动作检测器中的感兴趣区域(RoI)扩展到时域,在带有注意力机制的解码器中,通过场景上下文对捕获的时空动作执行者身份表征进行精炼。这两个设计使我们的EVAD在保持精度的同时具备高效性,该结论在三个基准数据集(即AVA、UCF101-24、JHMDB)上得到了验证。与原生ViT骨干网络相比,我们的EVAD将整体GFLOPs降低了43%,实时推理速度提升了40%,且无性能下降。此外,即使在相近计算成本下,使用更高分辨率输入的EVAD也能将性能提升1.1 mAP。代码可在https://github.com/MCG-NJU/EVAD获取。

英文摘要:

Streaming video clips with large-scale video tokens impede vision transformers (ViTs) for efficient recognition, especially in video action detection where sufficient spatiotemporal representations are required for precise actor identification. In this work, we propose an end-to-end framework for efficient video action detection (EVAD) based on vanilla ViTs. Our EVAD consists of two specialized designs for video action detection. First, we propose a spatiotemporal token dropout from a keyframe-centric perspective. In a video clip, we maintain all tokens from its keyframe, preserve tokens relevant to actor motions from other frames, and drop out the remaining tokens in this clip. Second, we refine scene context by leveraging remaining tokens for better recognizing actor identities. The region of interest (RoI) in our action detector is expanded into temporal domain. The captured spatiotemporal actor identity representations are refined via scene context in a decoder with the attention mechanism. These two designs make our EVAD efficient while maintaining accuracy, which is validated on three benchmark datasets (i.e., AVA, UCF101-24, JHMDB). Compared to the vanilla ViT backbone, our EVAD reduces the overall GFLOPs by 43% and improves real-time inference speed by 40% with no performance degradation. Moreover, even at similar computational costs, our EVAD can improve the performance by 1.1 mAP with higher resolution inputs. Code is available at https://github.com/MCG-NJU/EVAD.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑