Graph Attention Tracking
This work provides an incremental improvement for computer vision researchers working on object tracking by addressing specific limitations of existing Siamese network architectures.
This paper addresses the limitations of Siamese network-based trackers, which often include excessive background noise or loss of foreground information due to fixed target feature region sizes, and neglect of target structure. The authors propose a target-aware Siamese graph attention network that establishes part-to-part correspondence between target and search regions using a complete bipartite graph and propagates target information via a graph attention mechanism. The proposed method outperforms many state-of-the-art trackers on benchmarks like GOT-10k, UAV123, OTB-100, and LaSOT.
Siamese network based trackers formulate the visual tracking task as a similarity matching problem. Almost all popular Siamese trackers realize the similarity learning via convolutional feature cross-correlation between a target branch and a search branch. However, since the size of target feature region needs to be pre-fixed, these cross-correlation base methods suffer from either reserving much adverse background information or missing a great deal of foreground information. Moreover, the global matching between the target and search region also largely neglects the target structure and part-level information. In this paper, to solve the above issues, we propose a simple target-aware Siamese graph attention network for general object tracking. We propose to establish part-to-part correspondence between the target and the search region with a complete bipartite graph, and apply the graph attention mechanism to propagate target information from the template feature to the search feature. Further, instead of using the pre-fixed region cropping for template-feature-area selection, we investigate a target-aware area selection mechanism to fit the size and aspect ratio variations of different objects. Experiments on challenging benchmarks including GOT-10k, UAV123, OTB-100 and LaSOT demonstrate that the proposed SiamGAT outperforms many state-of-the-art trackers and achieves leading performance. Code is available at: https://git.io/SiamGAT