Research map / Detection and segmentation / Open-set and few-shot detection
HiMTok: Learning Hierarchical Mask Tokens for Image Segmentation with Large Multimodal Model
Tao Wang, Changxu Cheng, Lingfeng Wang, Senda Chen, Wuyue Zhao
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 hierarchical mask loss (HML) (ablation) | HiMTok-8B · RefCOCO val | cIoU: 79.2 -> 81.1 | Improved |
| add hierarchical mask loss (HML) (ablation) | HiMTok-8B · RefCOCO+ val | cIoU: 64.7 -> 77.1 | Improved |
| add hierarchical mask loss (HML) (ablation) | HiMTok-8B · RefCOCOg val | cIoU: 63.9 -> 75.8 | Improved |
| increase mask tokens from 16 to 32 (ablation) | HiMTok-8B · RefCOCO val | cIoU: 82.8% cIoU, additional 2.5% cIoU | Improved |
| increase mask token length before box prediction (ablation) | HiMTok-8B · RefCOCO val (REC) | Acc@0.9 | Improved |
| add box-to-mask information flow (ablation) | HiMTok-8B · RES | cIoU | No clear effect |
| add finetuning on RefCOCO/+/g training set | HiMTok-8B · RefCOCO val | cIoU: 81.1 -> 85.0 | Improved |
| add finetuned SAM refinement of de-tokenized mask | HiMTok-8B (ft) · RefCOCO val | cIoU: 85.0 -> 85.9 | Improved |
| add Mask Perception training data (ablation) | HiMTok-8B · Mask Perception test | accuracy: 63.1 -> 81.8 | Improved |
| add joint segmentation training (HiMTok) to InternVL2.5-8B | InternVL2.5-8B · RefCOCO val (REC) | Acc@0.5: 90.3 -> 92.9 | Improved |
| finetune on segmentation-only data | HiMTok-8B · MME count | score: 160 -> 113 | Hurt |
| add joint segmentation training (HiMTok) to InternVL2.5-8B | InternVL2.5-8B · MME scene | score: 154 -> 163 | Improved |
Later papers that compared with this method
1 later accepted paper reports HiMTok-8B as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Rethinking MLLM Itself as a Segmenter with a Single Segmentation Token (CVPR 2026 · better on 3 of 3 shared results)
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.
- PSALM — PSALM: Pixelwise Segmentation with Large Multi-modal Model (ECCV 2024) · better on 5 of 6 shared results
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
- Learning Hierarchical Image Segmentation For Recognition and By Recognition (ICLR 2024 · similarity 0.88)
- ALTo: Adaptive-Length Tokenizer for Autoregressive Mask Generation (NeurIPS 2025 · similarity 0.87)
- Subobject-level Image Tokenization (ICML 2025 · similarity 0.87)
- MaSS13K: A Matting-level Semantic Segmentation Benchmark (CVPR 2025 · similarity 0.87)
- Explore In-Context Segmentation via Latent Diffusion Models (AAAI 2025 · similarity 0.86)
- Pixels to Tokens: Token Space Efficient Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2026 · similarity 0.86)
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.
