Cascade-CLIP: Cascaded Vision-Language Embeddings Alignment for Zero-Shot Semantic Segmentation
This work improves zero-shot semantic segmentation for computer vision applications by enhancing feature alignment, though it is incremental as it builds on existing CLIP-based methods.
The paper tackles the problem of zero-shot semantic segmentation by addressing the neglect of intermediate layer visual features in CLIP-based methods, which contain rich object details but are hard to align with text embeddings due to large differences between layers. It introduces Cascade-CLIP, a framework using cascaded decoders to align multi-level features, achieving superior zero-shot performance on benchmarks like COCO-Stuff, Pascal-VOC, and Pascal-Context.
Pre-trained vision-language models, e.g., CLIP, have been successfully applied to zero-shot semantic segmentation. Existing CLIP-based approaches primarily utilize visual features from the last layer to align with text embeddings, while they neglect the crucial information in intermediate layers that contain rich object details. However, we find that directly aggregating the multi-level visual features weakens the zero-shot ability for novel classes. The large differences between the visual features from different layers make these features hard to align well with the text embeddings. We resolve this problem by introducing a series of independent decoders to align the multi-level visual features with the text embeddings in a cascaded way, forming a novel but simple framework named Cascade-CLIP. Our Cascade-CLIP is flexible and can be easily applied to existing zero-shot semantic segmentation methods. Experimental results show that our simple Cascade-CLIP achieves superior zero-shot performance on segmentation benchmarks, like COCO-Stuff, Pascal-VOC, and Pascal-Context. Our code is available at: https://github.com/HVision-NKU/Cascade-CLIP