Research map / Detection and segmentation / CNN features
Dual Path Networks
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 |
|---|---|---|---|
| Use a dual residual and densely connected path | DPN-92 · ImageNet-1k val, x224 | top-1 error rate (%): 21.2 -> 20.7 | Helped |
| Use a dual residual and densely connected path | DPN-98 · ImageNet-1k val, x224 | top-1 error rate (%): 20.4 -> 20.2 | Helped |
| Use a dual residual and densely connected path | DPN-92 · Places365-Standard val | top-1 accuracy (%): 56.60 -> 56.84 | Helped |
| Use a dual residual and densely connected path | Faster R-CNN with DPN-92 backbone · PASCAL VOC 2007 test | mAP (%): 80.1 -> 82.5 | Helped |
| Use a dual residual and densely connected path | DeepLab-ASPP-L with DPN-92 backbone · PASCAL VOC 2012 test | mIoU (%): 73.6 -> 74.8 | Helped |
| Add Mean-Max Pooling at test time (ablation) | DPN-92 · ImageNet-1k val | top-1 error rate (%): 19.34 -> 19.04 | Helped |
| Add Mean-Max Pooling at test time (ablation) | DPN-98 · ImageNet-1k val | top-1 error rate (%): 18.94 -> 18.72 | Helped |
| Add Mean-Max Pooling at test time (ablation) | DPN-131 · ImageNet-1k val | top-1 error rate (%): 18.62 -> 18.55 | Helped |
Closest papers by meaning
- Dual-stream Network for Visual Recognition (NeurIPS 2021 · similarity 0.87)
- Densely Connected Convolutional Networks (CVPR 2017 · similarity 0.85)
- Revisiting ResNets: Improved Training and Scaling Strategies (NeurIPS 2021 · similarity 0.85)
- Residual Alignment: Uncovering the Mechanisms of Residual Networks (NeurIPS 2023 · similarity 0.85)
- Dr2Net: Dynamic Reversible Dual-Residual Networks for Memory-Efficient Finetuning (CVPR 2024 · similarity 0.83)
- One-dimensional Path Convolution (ICML 2025 · similarity 0.83)
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.