Contrastive Transformation for Self-supervised Correspondence Learning
This work provides a more robust self-supervised method for visual correspondence, which is crucial for various computer vision tasks that rely on understanding object motion and relationships in videos.
This paper addresses self-supervised visual correspondence learning using unlabeled videos by simultaneously considering intra- and inter-video representation associations. The method achieves state-of-the-art performance in self-supervised correspondence and competes with fully-supervised methods on tasks like video object tracking and segmentation.
In this paper, we focus on the self-supervised learning of visual correspondence using unlabeled videos in the wild. Our method simultaneously considers intra- and inter-video representation associations for reliable correspondence estimation. The intra-video learning transforms the image contents across frames within a single video via the frame pair-wise affinity. To obtain the discriminative representation for instance-level separation, we go beyond the intra-video analysis and construct the inter-video affinity to facilitate the contrastive transformation across different videos. By forcing the transformation consistency between intra- and inter-video levels, the fine-grained correspondence associations are well preserved and the instance-level feature discrimination is effectively reinforced. Our simple framework outperforms the recent self-supervised correspondence methods on a range of visual tasks including video object tracking (VOT), video object segmentation (VOS), pose keypoint tracking, etc. It is worth mentioning that our method also surpasses the fully-supervised affinity representation (e.g., ResNet) and performs competitively against the recent fully-supervised algorithms designed for the specific tasks (e.g., VOT and VOS).