Coherent Temporal Synthesis for Incremental Action Segmentation

arXiv:2403.06102v1 Announce Type: new
Abstract: Data replay is a successful incremental learning technique for images. It prevents catastrophic forgetting by keeping a reservoir of previous data, original or synthesized, to ensure the model retains past knowledge while adapting to novel concepts. However, its application in the video domain is rudimentary, as it simply stores frame exemplars for action recognition. This paper presents the first exploration of video data replay techniques for incremental action segmentation, focusing on action temporal modeling. We …

abstract application arxiv catastrophic forgetting concepts cs.cv data domain however images incremental knowledge novel segmentation synthesis synthesized temporal type video