Semantic-SAM: Segment and Recognize Anything at Any Granularity

TL;DR

Semantic-SAM is a universal segmentation model that recognizes and segments objects at any granularity with semantic-awareness, trained on multiple datasets for rich semantic understanding.

Contribution

It introduces a multi-granularity training scheme and decoupled classification, enabling a single model to handle diverse segmentation tasks with semantic-awareness.

Findings

Achieves semantic-awareness and granularity-abundance.

Joint training on multiple datasets improves performance.

Demonstrates successful multi-level segmentation and recognition.

Abstract

In this paper, we introduce Semantic-SAM, a universal image segmentation model to enable segment and recognize anything at any desired granularity. Our model offers two key advantages: semantic-awareness and granularity-abundance. To achieve semantic-awareness, we consolidate multiple datasets across three granularities and introduce decoupled classification for objects and parts. This allows our model to capture rich semantic information. For the multi-granularity capability, we propose a multi-choice learning scheme during training, enabling each click to generate masks at multiple levels that correspond to multiple ground-truth masks. Notably, this work represents the first attempt to jointly train a model on SA-1B, generic, and part segmentation datasets. Experimental results and visualizations demonstrate that our model successfully achieves semantic-awareness and granularity-abundance. Furthermore, combining SA-1B training with other segmentation tasks, such as panoptic and part segmentation, leads to performance improvements. We will provide code and a demo for further exploration and evaluation.