Research map / Detection and segmentation / CNN features
ESPNetv2: A Light-Weight, Power Efficient, and General Purpose Convolutional Neural Network
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 |
|---|---|---|---|
| Use depth-wise dilated separable convolutions instead of depth-wise separable convolutions (ablation) | ESPNetv2 · ImageNet validation | top-1 accuracy: 66.5 -> 67.9 | Helped |
| Use depth-wise dilated separable convolutions instead of standard dilated convolutions (ablation) | ESPNetv2 · ImageNet validation | top-1 accuracy: 69.2 -> 67.9 | Hurt |
| Add hierarchical feature fusion (HFF) (ablation) | ESPNetv2 · ImageNet validation | top-1 accuracy: 58.94 -> 60.07 | Helped |
| Add a long-range shortcut connection to the input image (ablation) | ESPNetv2 · ImageNet validation | top-1 accuracy: 60.07 -> 61.20 | Helped |
| Use a cyclic rather than fixed learning schedule (ablation) | ESPNetv2 · ImageNet validation | top-1 accuracy: 61.20 -> 62.17 | Helped |
| Train for 300 epochs using the longer schedule (ablation) | ESPNetv2 · ImageNet validation | top-1 accuracy: 62.17 -> 66.10 | Helped |
Closest papers by meaning
- SCNN: A General Distribution Based Statistical Convolutional Neural Network with Application to Video Object Detection (AAAI 2019 · similarity 0.88)
- EfficientNetV2: Smaller Models and Faster Training (ICML 2021 · similarity 0.88)
- R-FCN: Object Detection via Region-based Fully Convolutional Networks (NeurIPS 2016 · similarity 0.87)
- Ubernet: Training a Universal Convolutional Neural Network for Low-, Mid-, and High-Level Vision Using Diverse Datasets and Limited Memory (CVPR 2017 · similarity 0.87)
- KernelWarehouse: Rethinking the Design of Dynamic Convolution (ICML 2024 · similarity 0.85)
- TGBFormer: Transformer-GraphFormer Blender Network for Video Object Detection (AAAI 2025 · similarity 0.85)
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.