Research map / Architectures and efficiency / CNN design
Context-Gated Convolution
Xudong Lin, Lin Ma, Wei Liu, Shih-Fu Chang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 context-gated convolution (CGC) | ResNet-50 · ImageNet val (Default) | Top-1: 76.16 -> 77.48 | Improved |
| add CGC under the Advanced training setting | ResNet-50 · ImageNet val (Advanced) | Top-1: 78.13 -> 79.54 | Improved |
| add CGC on top of CBAM | CBAM-ResNet-50 · ImageNet val | Top-1: 77.34 -> 77.68; 78.86 -> 79.74 | Improved |
| add CGC | ResNet-50 · ObjectNet | Top-1: 29.35 -> 31.53 | Improved |
| add CGC | ResNet-110 · CIFAR-10 | Top-1: 93.96 -> 94.86 | Improved |
| add CGC to TSN | TSN ResNet-50 · Something-Something v1 | Top-1: 19.00 -> 32.58 | Improved |
| add Non-local block to TSM | TSM ResNet-50 · Something-Something v1 | Top-1: 44.65 -> 43.91 | Hurt |
| add CGC to TSM | TSM ResNet-50 · Something-Something v1 | Top-1: 44.65 -> 46.00 | Improved |
| add CGC together with Non-local | TSM ResNet-50 · Kinetics | Top-1: 75.60 -> 76.40 | Improved |
| replace Lightweight Conv in encoder with CGC | LightConv · IWSLT'14 De-En | BLEU-4: 34.84 -> 35.21 | Improved |
| use both G(1) and G(2) via addition decomposition instead of only G(1) (ablation) | CGC · CIFAR-10 | Top-1: 94.53 -> 94.86 | Improved |
| use full linear channel interaction (g=1) instead of grouped (ablation) | CGC · CIFAR-10 | Top-1: 94.86 -> 94.97 | Improved |
Closest papers by meaning
- Contextual Convolutional Networks (ICLR 2023 · similarity 0.87)
- Context Reasoning Attention Network for Image Super-Resolution (ICCV 2021 · similarity 0.87)
- Convolutional Neural Networks with Gated Recurrent Connections (TPAMI 2021 · similarity 0.86)
- Learning to Predict Context-adaptive Convolution for Semantic Segmentation (ECCV 2020 · similarity 0.86)
- An Empirical Evaluation of Current Convolutional Architectures' Ability to Manage Nuisance Location and Scale Variability (CVPR 2016 · similarity 0.85)
- Self-Attention-Based Contextual Modulation Improves Neural System Identification (ICLR 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.
