Technical Report for ActivityNet Challenge 2022 -- Temporal Action Localization
This work addresses video analysis for researchers, but it is incremental as it builds on existing methods like Faster-TAD without introducing major innovations.
The authors tackled temporal action localization in untrimmed videos using the ActivityNet-1.3 dataset, achieving comparable performance to multi-step approaches by simplifying the pipeline with Faster-TAD and ensembling models.
In the task of temporal action localization of ActivityNet-1.3 datasets, we propose to locate the temporal boundaries of each action and predict action class in untrimmed videos. We first apply VideoSwinTransformer as feature extractor to extract different features. Then we apply a unified network following Faster-TAD to simultaneously obtain proposals and semantic labels. Last, we ensemble the results of different temporal action detection models which complement each other. Faster-TAD simplifies the pipeline of TAD and gets remarkable performance, obtaining comparable results as those of multi-step approaches.