Research map / Detection and segmentation / CNN features
RDSNet: A New Deep Architecture forReciprocal Object Detection and Instance 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 |
|---|---|---|---|
| Replace linear combination with the correlation module | YOLACT-style one-stage instance segmentation · COCO val2017 | mask AP: 29.9 -> 31.0 (+1.1) | Helped |
| Add expanded cropping during training and inference, with OHEM for background pixels (ablation) | RDSNet_s · COCO val2017 | mask AP: 29.9 -> 31.8 (+1.9) | Helped |
| Remove expanded cropping at inference (ablation) | RDSNet_s · COCO val2017 | mask AP: 31.0 -> 30.0 | Hurt |
| Expand boxes during training (ablation) | RDSNet_s · COCO val2017 | mask AP: 30.0 -> 30.7 | Helped |
| Add OHEM for background pixels (ablation) | RDSNet_s with expanded training crops · COCO val2017 | mask AP: 30.8 -> 31.6 | Helped |
| Add the mask branch to RetinaNet (ablation) | RetinaNet, ResNet-101, 800 input · COCO test-dev | box AP: 38.1 -> 39.4 | Helped |
| Add MBRM after the mask branch (ablation) | RDSNet, ResNet-101, 800 input · COCO test-dev | box AP: 39.4 -> 40.3 | Helped |
| Use the mask's minimum enclosing rectangle instead of the regressed box prior (ablation) | RDSNet · COCO val2017 | box AP: 35.9 -> 34.2 (-1.7) | Hurt |
| Add MBRM using the regression box as a prior (ablation) | RDSNet · COCO val2017 | box AP: 35.9 -> 37.2 (+1.3) | Helped |
Closest papers by meaning
- Instance-Aware Semantic Segmentation via Multi-Task Network Cascades (CVPR 2016 · similarity 0.89)
- Pixelwise Instance Segmentation With a Dynamically Instantiated Network (CVPR 2017 · similarity 0.89)
- Fully Convolutional Instance-Aware Semantic Segmentation (CVPR 2017 · similarity 0.88)
- Hybrid Task Cascade for Instance Segmentation (CVPR 2019 · similarity 0.88)
- BlendMask: Top-Down Meets Bottom-Up for Instance Segmentation (CVPR 2020 · similarity 0.88)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · 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.