Exploring Simple Open-Vocabulary Semantic Segmentation
This addresses the problem of pixel-level semantic segmentation with arbitrary text labels for computer vision researchers, offering a simpler and scalable baseline.
The paper tackles open-vocabulary semantic segmentation by introducing S-Seg, a model that achieves strong performance without relying on image-level VL models, ground truth masks, or custom grouping encoders, using pseudo-masks and language to train a MaskFormer from public datasets.
Open-vocabulary semantic segmentation models aim to accurately assign a semantic label to each pixel in an image from a set of arbitrary open-vocabulary texts. In order to learn such pixel-level alignment, current approaches typically rely on a combination of (i) image-level VL model (e.g. CLIP), (ii) ground truth masks, and (iii) custom grouping encoders. In this paper, we introduce S-Seg, a novel model that can achieve surprisingly strong performance without depending on any of the above elements. S-Seg leverages pseudo-mask and language to train a MaskFormer, and can be easily trained from publicly available image-text datasets. Contrary to prior works, our model directly trains for pixel-level features and language alignment. Once trained, S-Seg generalizes well to multiple testing datasets without requiring fine-tuning. In addition, S-Seg has the extra benefits of scalability with data and consistently improvement when augmented with self-training. We believe that our simple yet effective approach will serve as a solid baseline for future research.