Research map / Detection and segmentation / Open-set and few-shot detection
MTA-CLIP: Language-Guided Semantic Segmentation with Mask-Text Alignment
Anurag Das, Xinting Hu, Li Jiang, Bernt Schiele
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add text-enhanced mask feature learning (ablation) | Mask2Former (CLIP RN50) · ADE20k | mIoU: 47.4 -> 48.1 | Improved |
| add mask-to-text contrastive learning (ablation) | Mask2Former (CLIP RN50) with text-enhanced features · ADE20k | mIoU: 48.1 -> 48.6 | Improved |
| add mask-text prompt learning with MixNeg (ablation) | MTA-CLIP · ADE20k | mIoU: 48.6 -> 49.1 | Improved |
| let text embeddings and mask queries attend jointly in self-attention (ablation) | MTA-CLIP · ADE20k | mIoU: 48.6 -> 49.1 | Improved |
| use MixNeg instead of SeparateNeg for multiple prompts (ablation) | MTA-CLIP (K=3) · ADE20k | mIoU: 48.7 -> 49.1 | Improved |
| increase number of prompts K from 3 to 5 (ablation) | MTA-CLIP MixNeg · ADE20k | mIoU: 49.1 -> 48.4 | Hurt |
| replace pixel-text alignment with mask-text alignment | CLIP ResNet-50 segmentation · ADE20k | mIoU (SS): PPL 44.7 -> MTA-CLIP 49.1 | Improved |
| apply mask-text alignment to ImageNet-pretrained ViT-S backbone | Mask2Former · ADE20k | mIoU: 50.6 -> 51.4 | Improved |
| apply mask-text alignment to MiT-B4 backbone | SegFormer · ADE20k | mIoU: 50.3 -> 51.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SegFormer — SegFormer: A Topic Segmentation Model with Controllable Range of Attention (AAAI 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Unlocking Fine-Grained Perception in CLIP via Structurally-Aware Latent Masked Modeling (NeurIPS 2026 · similarity 0.89)
- Building Vision-Language Models on Solid Foundations with Masked Distillation (CVPR 2024 · similarity 0.89)
- Mask-Adapter: The Devil is in the Masks for Open-Vocabulary Segmentation (CVPR 2025 · similarity 0.88)
- Cascade-CLIP: Cascaded Vision-Language Embeddings Alignment for Zero-Shot Semantic Segmentation (ICML 2024 · similarity 0.88)
- DINOv2 Meets Text: A Unified Framework for Image- and Pixel-Level Vision-Language Alignment (CVPR 2025 · similarity 0.88)
- DPSeg: Dual-Prompt Cost Volume Learning for Open-Vocabulary Semantic Segmentation (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
