Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly Supervised Deep Detection Networks
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 |
|---|---|---|---|
| replace SSW proposals with EB proposals (ablation) | WSDDN S · VOC 2007 test | detection AP (%): 31.1 -> 31.5 | Helped |
| add EB Box Sc. (ablation) | WSDDN with models S, M, and L · VOC 2007 test | detection AP (%): 31.5 -> 33.4; 30.9 -> 32.7; 25.5 -> 30.4 | Helped |
| add spatial regularisation (ablation) | WSDDN with models S, M, and L · VOC 2007 test | detection AP (%): +1, +2 and +4 mAP points | Helped |
| use the deeper VGG-VD16 model instead of smaller WSDDN models (ablation) | WSDDN · VOC 2007 test | detection AP (%): 34.5; 34.9; 34.8 | Hurt |
| combine the three model scores as an ensemble (ablation) | WSDDN · VOC 2007 test | detection AP (%): 34.5; 34.9; 34.8 -> 39.3 | Helped |
| use WSDDN rather than the single-stream classification-detection baseline | PASCAL VOC test · PASCAL VOC test | detection mAP (%): 21.6 | Helped |
| fine-tune the pretrained VGG-F with WSDDN | VGG-CNN-F · VOC 2007 test | classification AP (%): 77.4 -> 85.3 | Helped |
| fine-tune the pretrained VGG-M-1024 with WSDDN | VGG-CNN-M-1024 · VOC 2007 test | classification AP (%): 79.9 -> 86.4 | Helped |
| use WSDDN L instead of VGG-VD16 | VGG-VD16 · VOC 2007 test | classification AP (%): 89.3 -> 89.7 | Helped |
| use WSDDN instead of Cinbis et al. | weakly supervised detector · VOC 2010 test | detection AP (%): 27.4 -> 36.2 | Helped |
Closest papers by meaning
- Weakly Supervised Cascaded Convolutional Networks (CVPR 2017 · similarity 0.93)
- Multi-Evidence Filtering and Fusion for Multi-Label Classification, Object Detection and Semantic Segmentation Based on Weakly Supervised Learning (CVPR 2018 · similarity 0.90)
- W2F: A Weakly-Supervised to Fully-Supervised Framework for Object Detection (CVPR 2018 · similarity 0.89)
- Multiple Instance Detection Network With Online Instance Classifier Refinement (CVPR 2017 · similarity 0.89)
- Instance-Aware, Context-Focused, and Memory-Efficient Weakly Supervised Object Detection (CVPR 2020 · similarity 0.88)
- Weakly Supervised Complementary Parts Models for Fine-Grained Image Classification From the Bottom Up (CVPR 2019 · 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.