OMG-Seg: Is One Model Good Enough For All Segmentation?

• DOI: 10.48550/arxiv.2401.10229
• Publication Date: 2024-01-01
• Authors: Xiangtai Li; Haobo Yuan; Wei Li; Henghui Ding; Size Wu; Wenwei Zhang; Yining Li; Chaoyu Chen; Chen Change Loy
• Keywords: Code (set theory)

In this work, we address various segmentation tasks, each traditionally tackled by distinct or partially unified models. We propose OMG-Seg, One Model that is Good enough to efficiently and effectively handle all the including image semantic, instance, panoptic segmentation, as well their video counterparts, open vocabulary settings, prompt-driven, interactive like SAM, object segmentation. To our knowledge, first model these tasks in one achieve satisfactory performance. show a transformer-based encoder-decoder architecture with task-specific queries outputs, can support over ten yet significantly reduce computational parameter overhead across datasets. rigorously evaluate inter-task influences correlations during co-training. Code models are available at https://github.com/lxtGH/OMG-Seg.