Long Short-Term Transformer for Online Action Detection
This addresses the problem of real-time action recognition in extended video sequences for applications like surveillance or video analysis, representing an incremental improvement over prior methods.
The paper tackles online action detection in long videos by introducing the Long Short-term TRansformer (LSTR), which uses a memory mechanism to model both long-term and short-term temporal information, achieving state-of-the-art performance on benchmarks like THUMOS'14, TVSeries, and HACS Segment.
We present Long Short-term TRansformer (LSTR), a temporal modeling algorithm for online action detection, which employs a long- and short-term memory mechanism to model prolonged sequence data. It consists of an LSTR encoder that dynamically leverages coarse-scale historical information from an extended temporal window (e.g., 2048 frames spanning of up to 8 minutes), together with an LSTR decoder that focuses on a short time window (e.g., 32 frames spanning 8 seconds) to model the fine-scale characteristics of the data. Compared to prior work, LSTR provides an effective and efficient method to model long videos with fewer heuristics, which is validated by extensive empirical analysis. LSTR achieves state-of-the-art performance on three standard online action detection benchmarks, THUMOS'14, TVSeries, and HACS Segment. Code has been made available at: https://xumingze0308.github.io/projects/lstr