Self-Disentanglement and Re-Composition for Cross-Domain Few-Shot Segmentation
This addresses a specific bottleneck in cross-domain few-shot segmentation for computer vision applications, representing an incremental improvement.
The paper tackles the entanglement problem in Cross-Domain Few-Shot Segmentation, where source-domain patterns are hard to transfer, by proposing a method to learn disentangled features and re-compose them, resulting in outperforming the state-of-the-art by 1.92% and 1.88% in average accuracy for 1-shot and 5-shot settings.
Cross-Domain Few-Shot Segmentation (CD-FSS) aims to transfer knowledge from a source-domain dataset to unseen target-domain datasets with limited annotations. Current methods typically compare the distance between training and testing samples for mask prediction. However, we find an entanglement problem exists in this widely adopted method, which tends to bind sourcedomain patterns together and make each of them hard to transfer. In this paper, we aim to address this problem for the CD-FSS task. We first find a natural decomposition of the ViT structure, based on which we delve into the entanglement problem for an interpretation. We find the decomposed ViT components are crossly compared between images in distance calculation, where the rational comparisons are entangled with those meaningless ones by their equal importance, leading to the entanglement problem. Based on this interpretation, we further propose to address the entanglement problem by learning to weigh for all comparisons of ViT components, which learn disentangled features and re-compose them for the CD-FSS task, benefiting both the generalization and finetuning. Experiments show that our model outperforms the state-of-the-art CD-FSS method by 1.92% and 1.88% in average accuracy under 1-shot and 5-shot settings, respectively.