Research map / Detection and segmentation / CNN features
Mask DINO: Towards a Unified Transformer-Based Framework for Object Detection and Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the mask prediction branch and unified detection-segmentation design | DINO · COCO val2017 instance segmentation, ResNet-50, 50 epochs | Mask AP: 43.7 -> 46.3 | Helped |
| Add mask-enhanced anchor box initialization (ablation) | Mask DINO · COCO val2017 object detection, ResNet-50, 12 epochs | Box AP: 44.5 -> 45.7 | Helped |
| Add unified query selection for masks (ablation) | Mask DINO · COCO val2017 instance segmentation, ResNet-50, 12 epochs | Mask AP: 40.3 -> 41.4 | Helped |
| Add unified query selection for masks (ablation) | Mask DINO · COCO val2017 object detection, ResNet-50, 12 epochs | Box AP: 43.6 -> 45.7 | Helped |
| Add unified denoising for masks (ablation) | Mask DINO · COCO val2017 instance segmentation, ResNet-50, 12 epochs | Mask AP: 40.3 -> 41.4 | Helped |
| Add unified denoising for masks (ablation) | Mask DINO · COCO val2017 object detection, ResNet-50, 12 epochs | Box AP: 44.4 -> 45.7 | Helped |
| Add hybrid matching using both box and mask costs (ablation) | Mask DINO · COCO val2017 instance segmentation, ResNet-50, 12 epochs | Mask AP: 40.5 -> 41.4 | Helped |
| Add hybrid matching using both box and mask costs (ablation) | Mask DINO · COCO val2017 object detection, ResNet-50, 12 epochs | Box AP: 44.9 -> 45.7 | Helped |
| Add all proposed components together (ablation) | Mask DINO · COCO val2017 instance segmentation, ResNet-50, 12 epochs | Mask AP: 38.5 -> 41.4 | Helped |
| Add all proposed components together (ablation) | Mask DINO · COCO val2017 object detection, ResNet-50, 12 epochs | Box AP: 41.7 -> 45.7 | Helped |
| Remove box loss and box matching for stuff categories (ablation) | Mask DINO panoptic segmentation · COCO val2017 panoptic segmentation, ResNet-50, 12 epochs | PQ: 47.9 -> 49.0 | Helped |
| Remove box loss and box matching for stuff categories (ablation) | Mask DINO panoptic segmentation · COCO val2017 panoptic segmentation, ResNet-50, 50 epochs | PQ: 52.7 -> 53.0 | Helped |
Closest papers by meaning
- DI-MaskDINO: A Joint Object Detection and Instance Segmentation Model (NeurIPS 2024 · similarity 0.88)
- Segment Anything, Even Occluded (CVPR 2025 · similarity 0.88)
- MaskDiff: Modeling Mask Distribution with Diffusion Probabilistic Model for Few-Shot Instance Segmentation (AAAI 2024 · similarity 0.87)
- EfficientSAM: Leveraged Masked Image Pretraining for Efficient Segment Anything (CVPR 2024 · similarity 0.87)
- PEM: Prototype-based Efficient MaskFormer for Image Segmentation (CVPR 2024 · similarity 0.87)
- Mask-Adapter: The Devil is in the Masks for Open-Vocabulary Segmentation (CVPR 2025 · 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.