Video Relation Detection via Tracklet based Visual Transformer
This addresses video relation understanding for multimedia analysis, but it is incremental as it builds on existing tracklet detection pipelines.
The paper tackled video relation detection by applying a tracklet-based visual Transformer to predict relations without pre-cutting, achieving superior performance that outperforms other methods by a large margin on the VRU Grand Challenge.
Video Visual Relation Detection (VidVRD), has received significant attention of our community over recent years. In this paper, we apply the state-of-the-art video object tracklet detection pipeline MEGA and deepSORT to generate tracklet proposals. Then we perform VidVRD in a tracklet-based manner without any pre-cutting operations. Specifically, we design a tracklet-based visual Transformer. It contains a temporal-aware decoder which performs feature interactions between the tracklets and learnable predicate query embeddings, and finally predicts the relations. Experimental results strongly demonstrate the superiority of our method, which outperforms other methods by a large margin on the Video Relation Understanding (VRU) Grand Challenge in ACM Multimedia 2021. Codes are released at https://github.com/Dawn-LX/VidVRD-tracklets.