Research map / Detection and segmentation / CNN features
HyperNet: Towards Accurate Region Proposal Generation and Joint Object Detection
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 |
|---|---|---|---|
| Fuse shallow, intermediate, and deep CNN features into Hyper Feature (ablation) | HyperNet · PASCAL VOC 2007, 100 proposals, IoU=0.5 | detection mAP: 64.2\% -> 65.9\% | Helped |
| Combine widely separated layers 1, 3, and 5 instead of adjacent layers 1, 2, and 3 (ablation) | AlexNet HyperNet · PASCAL VOC 2007, 100 proposals, IoU=0.5 | detection mAP: 63.8\% -> 65.9\% | Helped |
| Replace the full HyperNet with its speed-up architecture (ablation) | HyperNet · PASCAL VOC 2007 test, IoU=0.5 | mAP: 76.3 -> 74.8 | Hurt |
| Replace the full HyperNet with its speed-up architecture (ablation) | HyperNet · PASCAL VOC 2012 test | mAP: 71.4 -> 71.3 | No clear effect |
| Move the convolutional layer before ROI pooling (ablation) | HyperNet · PASCAL VOC 2007 test | total inference time (ms): 1140 -> 200 | Helped |
| Use VGG16 instead of AlexNet (ablation) | HyperNet · PASCAL VOC 2007 test, IoU=0.5 | mAP: 65.9\% -> 76.3\% | Helped |
| Use layers 3 and 5 rather than layer 5 alone (ablation) | AlexNet HyperNet · PASCAL VOC 2007, 100 proposals, IoU=0.5 | proposal recall: 94.98\% -> 95.00\% | No clear effect |
| Combine layers 1, 2, and 3 instead of layers 1, 3, and 5 (ablation) | AlexNet HyperNet · PASCAL VOC 2007, 100 proposals, IoU=0.5 | detection mAP: 65.9\% -> 63.8\% | Hurt |
| Combine layers 3, 4, and 5 instead of layers 1, 3, and 5 (ablation) | AlexNet HyperNet · PASCAL VOC 2007, 100 proposals, IoU=0.5 | detection mAP: 65.9\% -> 64.7\% | Hurt |
| Use only layer 1 instead of layers 1, 3, and 5 (ablation) | AlexNet HyperNet · PASCAL VOC 2007, 100 proposals, IoU=0.5 | proposal recall: 96.16\% -> 82.15\% | Hurt |
| Use only layer 3 instead of layers 1, 3, and 5 (ablation) | AlexNet HyperNet · PASCAL VOC 2007, 100 proposals, IoU=0.5 | proposal recall: 96.16\% -> 93.19\% | Hurt |
Closest papers by meaning
- You Reap What You Sow: Using Videos to Generate High Precision Object Proposals for Weakly-Supervised Object Detection (CVPR 2019 · similarity 0.89)
- R-FCN: Object Detection via Region-based Fully Convolutional Networks (NeurIPS 2016 · similarity 0.89)
- RON: Reverse Connection With Objectness Prior Networks for Object Detection (CVPR 2017 · similarity 0.88)
- Sylph: A Hypernetwork Framework for Incremental Few-Shot Object Detection (CVPR 2022 · similarity 0.88)
- LocNet: Improving Localization Accuracy for Object Detection (CVPR 2016 · similarity 0.88)
- Recognize Any Regions (NeurIPS 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.