PaintSeg: Training-free Segmentation via Painting
This provides a training-free solution for unsupervised segmentation, which is incremental as it builds on existing generative models but offers a novel method for a known bottleneck in segmentation tasks.
The paper tackles unsupervised object segmentation without training by introducing PaintSeg, which uses an adversarial masked contrastive painting process with inpainting and outpainting steps to refine segmentation masks, and it outperforms existing methods in tasks like coarse mask, box, and point-prompt segmentation.
The paper introduces PaintSeg, a new unsupervised method for segmenting objects without any training. We propose an adversarial masked contrastive painting (AMCP) process, which creates a contrast between the original image and a painted image in which a masked area is painted using off-the-shelf generative models. During the painting process, inpainting and outpainting are alternated, with the former masking the foreground and filling in the background, and the latter masking the background while recovering the missing part of the foreground object. Inpainting and outpainting, also referred to as I-step and O-step, allow our method to gradually advance the target segmentation mask toward the ground truth without supervision or training. PaintSeg can be configured to work with a variety of prompts, e.g. coarse masks, boxes, scribbles, and points. Our experimental results demonstrate that PaintSeg outperforms existing approaches in coarse mask-prompt, box-prompt, and point-prompt segmentation tasks, providing a training-free solution suitable for unsupervised segmentation.