TAG: Guidance-free Open-Vocabulary Semantic Segmentation

TL;DR

The paper introduces TAG, a novel guidance-free open-vocabulary semantic segmentation method that leverages pre-trained models to segment images without additional training or dense annotations, achieving state-of-the-art results.

Contribution

TAG is the first approach to perform open-vocabulary segmentation without guidance or training, using pre-trained models and external class label retrieval.

Findings

Achieves +15.3 mIoU improvement on PascalVOC

State-of-the-art results on PascalContext and ADE20K

Operates without class name guidance or additional training

Abstract

Semantic segmentation is a crucial task in computer vision, where each pixel in an image is classified into a category. However, traditional methods face significant challenges, including the need for pixel-level annotations and extensive training. Furthermore, because supervised learning uses a limited set of predefined categories, models typically struggle with rare classes and cannot recognize new ones. Unsupervised and open-vocabulary segmentation, proposed to tackle these issues, faces challenges, including the inability to assign specific class labels to clusters and the necessity of user-provided text queries for guidance. In this context, we propose a novel approach, TAG which achieves Training, Annotation, and Guidance-free open-vocabulary semantic segmentation. TAG utilizes pre-trained models such as CLIP and DINO to segment images into meaningful categories without additional training or dense annotations. It retrieves class labels from an external database, providing flexibility to adapt to new scenarios. Our TAG achieves state-of-the-art results on PascalVOC, PascalContext and ADE20K for open-vocabulary segmentation without given class names, i.e. improvement of +15.3 mIoU on PascalVOC. All code and data will be released at https://github.com/Valkyrja3607/TAG.