Research map / Architectures and efficiency / CNN design
Xception: Deep Learning With Depthwise Separable Convolutions
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 Inception modules with depthwise separable convolutions in Xception | Inception V3 · ImageNet validation | Top-1 accuracy: 0.782 -> 0.790 | Helped |
| replace Inception modules with depthwise separable convolutions in Xception | Inception V3 · ImageNet validation | Top-5 accuracy: 0.941 -> 0.945 | Helped |
| replace Inception modules with depthwise separable convolutions in Xception | Inception V3 - no FC layers · FastEval14k | MAP@100: 6.36 -> 6.70 | Helped |
| add two fully-connected layers of 4096 units each (ablation) | Xception - no FC layers · FastEval14k | MAP@100: 6.70 -> 6.78 | Helped |
| add an intermediate ReLU or ELU between depthwise and pointwise operations (ablation) | Xception · ImageNet | classification performance | Hurt |
| remove residual connections (ablation) | Xception · ImageNet | classification performance | Hurt |
| remove residual connections (ablation) | Xception · ImageNet | convergence speed | Hurt |
Closest papers by meaning
- Rethinking the Inception Architecture for Computer Vision (CVPR 2016 · similarity 0.87)
- Depthwise Separable Convolutions for Neural Machine Translation (ICLR 2018 · similarity 0.86)
- InceptionNeXt: When Inception Meets ConvNeXt (CVPR 2024 · similarity 0.85)
- Unveiling the Unseen: Identifiable Clusters in Trained Depthwise Convolutional Kernels (ICLR 2024 · similarity 0.83)
- Model LEGO: Creating Models Like Disassembling and Assembling Building Blocks (NeurIPS 2024 · similarity 0.83)
- DD-CAM: Minimal Sufficient Explanations for Vision Models Using Delta Debugging (NeurIPS 2026 · similarity 0.82)
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.