Learnable Graph Matching: A Practical Paradigm for Data Association
This addresses data association issues in computer vision tasks like tracking and matching, offering a general solution that integrates deep learning with optimization, though it is incremental in combining existing techniques.
The paper tackles data association problems in computer vision by proposing a learnable graph matching method that models intra-view relationships as graphs and relaxes the matching into differentiable optimization, achieving state-of-the-art performance on multiple object tracking datasets and outperforming others in image matching on ScanNet.
Data association is at the core of many computer vision tasks, e.g., multiple object tracking, image matching, and point cloud registration. however, current data association solutions have some defects: they mostly ignore the intra-view context information; besides, they either train deep association models in an end-to-end way and hardly utilize the advantage of optimization-based assignment methods, or only use an off-the-shelf neural network to extract features. In this paper, we propose a general learnable graph matching method to address these issues. Especially, we model the intra-view relationships as an undirected graph. Then data association turns into a general graph matching problem between graphs. Furthermore, to make optimization end-to-end differentiable, we relax the original graph matching problem into continuous quadratic programming and then incorporate training into a deep graph neural network with KKT conditions and implicit function theorem. In MOT task, our method achieves state-of-the-art performance on several MOT datasets. For image matching, our method outperforms state-of-the-art methods on a popular indoor dataset, ScanNet. For point cloud registration, we also achieve competitive results. Code will be available at https://github.com/jiaweihe1996/GMTracker.