Research map / Detection and segmentation / CNN features
Instance-Aware Semantic Segmentation via Multi-Task Network Cascades
Check 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 |
|---|---|---|---|
| Share convolutional features without end-to-end training (ablation) | MNC with VGG-16 · PASCAL VOC 2012 val | mAP^r@0.5: 60.2 -> 60.5 | No clear effect |
| Train the 3-stage cascade end-to-end instead of using step-by-step training (ablation) | MNC with VGG-16 · PASCAL VOC 2012 val | mAP^r@0.5: 60.5 -> 62.6 | Helped |
| Train a 5-stage cascade rather than a 3-stage cascade (ablation) | MNC with VGG-16 · PASCAL VOC 2012 val | mAP^r@0.5: 62.6 -> 63.5 | Helped |
| Train the 3-stage cascade end-to-end instead of using step-by-step training (ablation) | MNC with ZF net · PASCAL VOC 2012 val | mAP^r@0.5: 52.2 -> 53.5 | Helped |
| Train a 5-stage cascade rather than a 3-stage cascade (ablation) | MNC with ZF net · PASCAL VOC 2012 val | mAP^r@0.5: 53.5 -> 54.0 | Helped |
| Use ResNet-101 instead of VGG-16 | MNC · MS COCO test-dev | mAP@[.5:.95]: 19.5 -> 24.6 | Helped |
| Apply mask voting to merge similar instance masks | MNC post-processing · PASCAL VOC 2012 val | mAP^r: ∼1% | Helped |
| Use MNC instead of the strongest listed prior segmentation baseline | Instance-aware semantic segmentation · PASCAL VOC 2012 val | mAP^r@0.5: 60.7 -> 63.5 | Helped |
| Use MNC instead of the strongest listed prior segmentation baseline | Instance-aware semantic segmentation · PASCAL VOC 2012 val | mAP^r@0.7: 40.4 -> 41.5 | Helped |
Closest papers by meaning
- Not All Pixels Are Equal: Difficulty-Aware Semantic Segmentation via Deep Layer Cascade (CVPR 2017 · similarity 0.92)
- Pixelwise Instance Segmentation With a Dynamically Instantiated Network (CVPR 2017 · similarity 0.92)
- Hybrid Task Cascade for Instance Segmentation (CVPR 2019 · similarity 0.92)
- Fully Convolutional Instance-Aware Semantic Segmentation (CVPR 2017 · similarity 0.92)
- Semantic-aware SAM for Point-Prompted Instance Segmentation (CVPR 2024 · similarity 0.90)
- P2Seg: Pointly-supervised Segmentation via Mutual Distillation (ICLR 2024 · similarity 0.88)
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.