Rethinking Alignment and Uniformity in Unsupervised Image Semantic Segmentation

Unsupervised image segmentation aims to match low-level visual features with semantic-level representations without outer supervision. In this paper, we address the critical properties from view of feature alignments and uniformity for UISS models. We also make a comparison between image-wise representation learning. Based on analysis, argue that existing MI-based methods in suffer collapse. By this, proposed robust network called Semantic Attention Network(SAN), which new module Attention(SEAT) is generate pixel-wise semantic dynamically. Experimental results multiple benchmarks show our unsupervised framework specializes catching representations, outperforms all unpretrained even several pretrained methods.