Research map / Detection and segmentation / Weakly supervised detection and segmentation
Multiple Instance Detection Network With Online Instance Classifier Refinement
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 |
|---|---|---|---|
| Add online instance classifier refinement using spatially overlapping proposals (ablation) | Basic MIDN · VOC 2007 | mAP: 29.5 to 35.6 | Helped |
| Add online instance classifier refinement using spatially overlapping proposals (ablation) | Basic MIDN · VOC 2007 | CorLoc: 49.9 to 56.0 | Helped |
| Use weighted refined-classifier loss instead of unweighted loss (ablation) | OICR · VOC 2007 | mAP: 32.8 -> 37.9 | Helped |
| Use weighted refined-classifier loss instead of unweighted loss (ablation) | OICR · VOC 2007 | CorLoc: 50.6 -> 57.3 | Helped |
| Raise the proposal-overlap IoU threshold from 0.5 to 0.6 (ablation) | OICR · VOC 2007 | mAP: 37.9 to 37.8 | Hurt |
| Raise the proposal-overlap IoU threshold from 0.5 to 0.6 (ablation) | OICR · VOC 2007 | CorLoc: 57.3 | No clear effect |
| Use three refinement stages rather than two (ablation) | OICR · VOC 2007 | mAP | No clear effect |
| Train the refinement stages jointly with shared representations using OICR rather than separate alternative training (ablation) | Multi-stage instance classifier refinement · VOC 2007 | mAP | Helped |
| Combine VGG_M and VGG16 OICR scores | OICR · VOC 2007 test | mAP: 41.2 / 37.9 -> 42.0 | Helped |
| Train Fast R-CNN on top-scoring OICR proposals used as pseudo ground truths | OICR-Ens. · VOC 2007 test | mAP: 42.0 -> 47.0 | Helped |
| Train Fast R-CNN on top-scoring OICR proposals used as pseudo ground truths | OICR-Ens. · VOC 2012 test | mAP: 38.2 -> 42.5 | Helped |
Closest papers by meaning
- Instance-Aware, Context-Focused, and Memory-Efficient Weakly Supervised Object Detection (CVPR 2020 · similarity 0.91)
- Object Instance Mining for Weakly Supervised Object Detection (AAAI 2020 · similarity 0.90)
- C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection (CVPR 2019 · similarity 0.90)
- W2F: A Weakly-Supervised to Fully-Supervised Framework for Object Detection (CVPR 2018 · similarity 0.89)
- Weakly Supervised Deep Detection Networks (CVPR 2016 · similarity 0.89)
- Instance Mining with Class Feature Banks for Weakly Supervised Object Detection (AAAI 2021 · similarity 0.89)
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.