Research map / Vision-language and multimodal / CLIP and zero-shot
Derm1M: A Million-scale Vision-Language Dataset Aligned with Clinical Ontology Knowledge for Dermatology
Siyuan Yan, Ming Hu, Yiwen Jiang, Xieji Li, Hao Fei, Philipp Tschandl, Harald Kittler, Zongyuan Ge
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 |
|---|---|---|---|
| train CLIP-style model on Derm1M (DermLIP ViT-B16/GPT77) vs BiomedCLIP | DermLIP · Zero-shot classification average | accuracy: .4408 -> .5607 | Improved |
| use PanDerm-B vision and PubMedBert text encoders instead of ViT-B16/GPT77 (ablation) | DermLIP · Zero-shot classification average | accuracy: .5607 -> .5879 | Improved |
| use PanDerm-B/PMB256 instead of ViT-B16/GPT77 (ablation) | DermLIP · Zero-shot Fitzpatrick17K | accuracy: .2278 -> .3162 | Improved |
| use PanDerm-B/PMB256 instead of ViT-B16/GPT77 (ablation) | DermLIP · Zero-shot HAM10000 | accuracy: .6820 -> .6274 | Hurt |
| use longer text context (PMB256) with generic ViT-B16 instead of GPT77 (ablation) | DermLIP · Zero-shot classification average | accuracy: .5607 -> .5558 | Mixed |
| use PanDerm-B/PMB256 instead of ViT-B16/GPT77 (ablation) | DermLIP · Linear eval average, 1% labels | accuracy: .570 -> .586 | Improved |
| use PanDerm-B/PMB256 instead of ViT-B16/GPT77 (ablation) | DermLIP · Linear eval average, 10% labels | accuracy: .691 -> .696 | Improved |
| pretrain on Derm1M vs MONET | DermLIP PanDerm-B/PMB256 · Linear eval average, 1% labels | accuracy: .530 -> .586 | Improved |
| use PanDerm-B/PMB256 instead of ViT-B16/GPT77 (ablation) | DermLIP · Retrieval Derm1M holdout I2T | Recall@10: .4069 -> .5998 | Improved |
| use PanDerm-B/PMB256 instead of ViT-B16/GPT77 (ablation) | DermLIP · Retrieval SkinCAP I2T | Recall@10: .1567 -> .2018 | Improved |
| DermLIP vs MONET on concept identification | DermLIP ViT-B16/GPT77 · Derm7pt | AUROC: 68.8% vs. 68.9% | No clear effect |
| DermLIP vs MONET on concept identification | DermLIP ViT-B16/GPT77 · SkinCon + Derm7pt | AUROC: 73.1% | 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.
- CoCa — COCA: COllaborative CAusal Regularization for Audio-Visual Question Answering (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
- DermaCon-IN: A Multiconcept-Annotated Dermatological Image Dataset of Indian Skin Disorders for Clinical AI Research (NeurIPS 2025 · similarity 0.91)
- BIOMEDICA: An Open Biomedical Image-Caption Archive, Dataset, and Vision-Language Models Derived from Scientific Literature (CVPR 2025 · similarity 0.88)
- Doctor Approved: Generating Medically Accurate Skin Disease Images through AI-Expert Feedback (NeurIPS 2025 · similarity 0.87)
- MICA: Towards Explainable Skin Lesion Diagnosis via Multi-Level Image-Concept Alignment (AAAI 2024 · similarity 0.85)
- Medical Manifestation-Aware De-Identification (AAAI 2025 · similarity 0.84)
- MedTrinity-25M: A Large-scale Multimodal Dataset with Multigranular Annotations for Medicine (ICLR 2025 · similarity 0.83)
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.
