Research map / Detection and segmentation / Open-set and few-shot detection
DAMEX: Dataset-aware Mixture-of-Experts for visual understanding of mixture-of-datasets
Yash Jain, Harkirat Behl, Zsolt Kira, Vibhav Vineet
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 |
|---|---|---|---|
| route tokens to dataset-mapped experts | DINO trained on UODB · UODB | Mean AP: +2.0 AP | Improved |
| use dataset-aware routing rather than vanilla MoE routing | DINO-MoE on UODB · UODB | Mean AP: +1.3 AP | Improved |
| route dataset-specific features to specialized experts | DINO trained on DOTA and DeepLesion · DOTA, DeepLesion | AP | Improved |
| use dataset-aware routing | DINO trained on KITTI · UODB | AP | Hurt |
| use DAMEX rather than non-MoE DINO | KITTI and Kitchen mixture with 50-shot KITTI · KITTI | AP: 14.6 -> 16.9 | Improved |
| use DAMEX rather than vanilla MoE | KITTI and Kitchen mixture with 100-shot KITTI · KITTI | AP: 20.9 -> 23.5 | Improved |
| use DAMEX rather than vanilla MoE | KITTI and Kitchen mixture with 1000-shot KITTI · KITTI | AP: 45.2 -> 46.1 | Improved |
| use DAMEX rather than vanilla MoE | Watercolor and Comic mixture · Watercolor, Comic | AP: +2.9 AP | Improved |
| use DAMEX rather than vanilla MoE | KITTI and LISA mixture · KITTI, LISA | AP: +1.7 AP on KITTI; +2.9 AP on LISA | Improved |
| use human-prior dataset-to-expert mapping instead of random mapping (ablation) | DAMEX on UODB · UODB | Mean AP: 43.7 -> 44.3 | Improved |
| use 4 experts rather than 2 experts (ablation) | DAMEX on KITTI, DOTA, VOC, and Clipart · KITTI, DOTA, VOC, Clipart | Mean AP: 48.6 -> 49.7 | Improved |
| use 8 experts rather than 4 experts (ablation) | DAMEX on KITTI, DOTA, VOC, and Clipart · KITTI, DOTA, VOC, Clipart | Mean AP: 49.7 -> 48.7 | Hurt |
Later papers that compared with this method
1 later accepted paper reports DAMEX as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- COME: Dual Structure-Semantic Learning with Collaborative MoE for Universal Lesion Detection Across Heterogeneous Ultrasound Datasets (ICCV 2025 · 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
- Multilinear Mixture of Experts: Scalable Expert Specialization through Factorization (NeurIPS 2024 · similarity 0.87)
- Efficient Data Driven Mixture-of-Expert Extraction from Trained Networks (CVPR 2025 · similarity 0.87)
- Mixture of Nested Experts: Adaptive Processing of Visual Tokens (NeurIPS 2024 · similarity 0.86)
- MoDE: CLIP Data Experts via Clustering (CVPR 2024 · similarity 0.86)
- Mixture of Experts as Representation Learner for Deep Multi-View Clustering (AAAI 2025 · similarity 0.85)
- MoDE: A Mixture-of-Experts Model with Mutual Distillation among the Experts (AAAI 2024 · similarity 0.85)
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.
