Research map / Vision-language and multimodal / Cross-modal learning
FAME-ViL: Multi-Tasking Vision-Language Model for Heterogeneous Fashion Tasks
Xiao Han, Xiatian Zhu, Licheng Yu, Li Zhang, Yi-Zhe Song, Tao Xiang
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 one multi-task FAME-ViL model instead of its independently trained single-task version | FAME-ViL · FashionGen XMR (random 100) | Mean R@K: 82.61 -> 83.14 | Improved |
| Use one multi-task FAME-ViL model instead of its independently trained single-task version | FAME-ViL · FashionIQ TGIR | Mean R@K: 55.36 -> 58.29 | Improved |
| Add TSA to the MTL model (ablation) | CLIP-based MTL · Four-task average | Average per-task relative performance: -5.59 -> +0.71 | Improved |
| Add XAA alone to the MTL model (ablation) | CLIP-based MTL · Four-task average | Average per-task relative performance: -5.59 -> -0.72 | Hurt |
| Add both TSA and XAA to MTL (ablation) | CLIP-based MTL · Four-task average | Average per-task relative performance: -5.59 -> +3.56 | Improved |
| Add multi-teacher distillation to MTL with TSA and XAA (ablation) | CLIP-based MTL · Four-task average | Average per-task relative performance: +3.56 -> +5.50 | Improved |
| Replace size-proportional task sampling with uniform sampling (ablation) | FAME-ViL training · Four-task average | Average per-task relative performance: +5.50 -> +3.99 | Hurt |
| Replace size-proportional task sampling with round-robin sampling (ablation) | FAME-ViL training · Four-task average | Average per-task relative performance: +5.50 -> +4.00 | Hurt |
| Use IAS instead of MTD for task balancing (ablation) | Base MTL · Four-task average | Average per-task relative performance: +5.50 -> +2.74 | Hurt |
| Add MTD to MTL with IAS (ablation) | Base MTL + IAS · Four-task average | Average per-task relative performance: +2.74 -> +5.30 | Improved |
| Use IMTLG instead of MTD for task balancing (ablation) | Base MTL · Four-task average | Average per-task relative performance: +5.50 -> -6.33 | Hurt |
| Add MTD to MTL with IMTLG (ablation) | Base MTL + IMTLG · Four-task average | Average per-task relative performance: -6.33 -> +0.44 | Improved |
Later papers that compared with this method
2 later accepted papers report FAME-ViL as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Heterogeneous Uncertainty-Guided Composed Image Retrieval with Fine-Grained Probabilistic Learning (AAAI 2026 · better on 1 of 1 shared result)
- ENCODER: Entity Mining and Modification Relation Binding for Composed Image Retrieval (AAAI 2025 · better on 2 of 2 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.
- FashionViL — FashionViL: Fashion-Focused Vision-and-Language Representation Learning (ECCV 2022) · better on 2 of 2 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
- Easy Regional Contrastive Learning of Expressive Fashion Representations (NeurIPS 2024 · similarity 0.90)
- Seeing the Abstract: Translating the Abstract Language for Vision Language Models (CVPR 2025 · similarity 0.88)
- VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks (ICLR 2025 · similarity 0.86)
- VL-ICL Bench: The Devil in the Details of Multimodal In-Context Learning (ICLR 2025 · similarity 0.86)
- FashionERN: Enhance-and-Refine Network for Composed Fashion Image Retrieval (AAAI 2024 · similarity 0.86)
- Benchmarking Large Vision-Language Models on Fine-Grained Image Tasks: A Comprehensive Evaluation (ICLR 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.
