In this work, we investigate the pixel-level semantic aggregation in self-supervised ViT pre-trained models as image Segmentation and propose the Adaptive Conceptualization approach for USS, termed ACSeg  ...  prediction tasks, e.g., unsupervised semantic segmentation (USS).  ...  We thus propose the Adaptive Conceptualization for unsupervised semantic Segmentation (ACSeg), a framework that finds these underlying concepts adaptively for each image and achieves USS by classifying  ...