Learning Hierarchical Image Segmentation For Recognition and By Recognition

arXiv:2210.00314v4 Announce Type: replace-cross
Abstract: Large vision and language models learned directly through image-text associations often lack detailed visual substantiation, whereas image segmentation tasks are treated separately from recognition, supervisedly learned without interconnections. Our key observation is that, while an image can be recognized in multiple ways, each has a consistent part-and-whole visual organization. Segmentation thus should be treated not as an end task to be mastered through supervised learning, but as an internal process that evolves with and supports …

abstract arxiv consistent cs.ai cs.cv cs.lg hierarchical image key language language models multiple observation recognition segmentation tasks text through type vision visual while