Research map / Label-efficient and robust learning / Recognition and long-tail
Joint Acne Image Grading and Counting via Label Distribution Learning
Xiaoping Wu, Ni Wen, Jie Liang, Yu-Kun Lai, Dongyu She, Ming-Ming Cheng, Jufeng Yang
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 |
|---|---|---|---|
| use label distribution instead of single label for grading (ablation) | ResNet-50 grading · ACNE04 | accuracy: 78.42 -> 79.31 | Improved |
| use label distribution of lesion number instead of single label for counting (ablation) | ResNet-50 counting · ACNE04 | MAE: 4.16 -> 3.24 | Improved |
| use lesion-count label distribution, with counts converted to grade (ablation) | ResNet-50 counting · ACNE04 | accuracy: 75.69 -> 81.09 | Improved |
| use dynamically generated grading label distribution instead of standard LDL (ablation) | ResNet-50 grading · ACNE04 | accuracy: 79.31 -> 82.05 | Improved |
| add joint grading and counting multi-task learning (ablation) | Ours(G) · ACNE04 | accuracy: 82.05 -> 82.53 | Improved |
| add joint grading and counting multi-task learning (ablation) | LDL(C) · ACNE04 | MAE: 3.24 -> 3.01 | Improved |
| add Lcnt2cls loss and averaging of grading and counting predictions (ablation) | Ours(G+C) · ACNE04 | accuracy: 82.53 -> 84.11 | Improved |
| add Lcnt2cls loss and averaging of grading and counting predictions (ablation) | Ours(G+C) · ACNE04 | MAE: 3.01 -> 2.93 | Improved |
| increase lambda from 0.1 up to 0.6 (ablation) | Ours · ACNE04 | accuracy and MAE | Mixed |
| replace ResNet-50 with the proposed method | ResNet-50 · ACNE04 | accuracy: 78.42 -> 84.11 | Improved |
| use proposed method instead of Faster R-CNN based counting | F-RCNN · ACNE04 | MAE: 6.70 -> 2.93 | Improved |
| use deep features instead of hand-crafted features | SVM classifiers · ACNE04 | accuracy: 48.22 -> 78.42 | Improved |
Closest papers by meaning
- BiasMitigateGan: SYNTHESIZING FAIR TRAINING DATA FOR DERMATOLOGY AI USING DIFFUSION MODELS (IJCV 2025 · similarity 0.83)
- DermaCon-IN: A Multiconcept-Annotated Dermatological Image Dataset of Indian Skin Disorders for Clinical AI Research (NeurIPS 2025 · similarity 0.80)
- Doctor Approved: Generating Medically Accurate Skin Disease Images through AI-Expert Feedback (NeurIPS 2025 · similarity 0.80)
- Approximately Correct Label Distribution Learning (ICML 2025 · similarity 0.79)
- MICA: Towards Explainable Skin Lesion Diagnosis via Multi-Level Image-Concept Alignment (AAAI 2024 · similarity 0.79)
- Samba: Severity-aware Recurrent Modeling for Cross-domain Medical Image Grading (NeurIPS 2024 · similarity 0.78)
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.
