SemiVL: Semi-Supervised Semantic Segmentation with Vision-Language Guidance
About
In semi-supervised semantic segmentation, a model is trained with a limited number of labeled images along with a large corpus of unlabeled images to reduce the high annotation effort. While previous methods are able to learn good segmentation boundaries, they are prone to confuse classes with similar visual appearance due to the limited supervision. On the other hand, vision-language models (VLMs) are able to learn diverse semantic knowledge from image-caption datasets but produce noisy segmentation due to the image-level training. In SemiVL, we propose to integrate rich priors from VLM pre-training into semi-supervised semantic segmentation to learn better semantic decision boundaries. To adapt the VLM from global to local reasoning, we introduce a spatial fine-tuning strategy for label-efficient learning. Further, we design a language-guided decoder to jointly reason over vision and language. Finally, we propose to handle inherent ambiguities in class labels by providing the model with language guidance in the form of class definitions. We evaluate SemiVL on 4 semantic segmentation datasets, where it significantly outperforms previous semi-supervised methods. For instance, SemiVL improves the state-of-the-art by +13.5 mIoU on COCO with 232 annotated images and by +6.1 mIoU on Pascal VOC with 92 labels. Project page: https://github.com/google-research/semivl
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Change Detection | LEVIR-CD (test) | -- | 357 | |
| Semantic segmentation | Pascal VOC (Original set) | mIoU87.3 | 105 | |
| Semantic segmentation | Cityscapes 1/4 (744 labels) | mIoU80.3 | 80 | |
| Semantic segmentation | Cityscapes 1/16 (186 labeled samples) | mIoU77.9 | 68 | |
| Semantic segmentation | CITYSCAPES 1/8 labeled samples 372 labels (val) | mIoU79.4 | 65 | |
| Semantic segmentation | Pascal VOC 1/16 labeled 2012 (train) | mIoU84 | 53 | |
| Semantic segmentation | Pascal VOC Original protocol 92 labeled images | mIoU84 | 48 | |
| Semantic segmentation | Pascal VOC Original protocol 1464 labeled images | mIoU87.3 | 36 | |
| Semantic segmentation | Pascal VOC 183 labeled images (Original protocol) | mIoU85.6 | 34 | |
| Semantic segmentation | Pascal VOC 366 labeled images (Original protocol) | mIoU86 | 34 |