Research map / Label-efficient and robust learning / Domain adaptation and generalization
Depthwise Convolution Is All You Need for Learning Multiple Visual Domains
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 |
|---|---|---|---|
| Share pointwise convolution across domains and use domain-specific depthwise filters instead of sharing depthwise convolution | ResNet-26 multi-domain model · Visual Decathlon Challenge | S: 3234 -> 3507 | Helped |
| Replace standard convolution with depthwise separable convolution | ResNet-26 · ImageNet | Top-1 accuracy: 60.32 -> 63.99 | Helped |
| Add soft sharing of trained depthwise filters in early layers (ablation) | Proposed multi-domain model · Visual Decathlon Challenge | S: 3507 -> 3102 | Hurt |
| Add soft sharing of trained depthwise filters in middle layers (ablation) | Proposed multi-domain model · Visual Decathlon Challenge | S: 3507 -> 3086 | Hurt |
| Add soft sharing of trained depthwise filters in late layers (ablation) | Proposed multi-domain model · Visual Decathlon Challenge | S: 3507 -> 3303 | Hurt |
| Add soft sharing of trained depthwise filters in early layers (ablation) | Proposed multi-domain model · DTD | Top-1 accuracy: 55.48 -> 55.75 | Helped |
| Add soft sharing of trained depthwise filters in early layers (ablation) | Proposed multi-domain model · SVHN | Top-1 accuracy: 96.16 -> 96.18 | Helped |
| Use late-layer rather than early-layer soft sharing (ablation) | Proposed multi-domain model · Visual Decathlon Challenge | S: 3102 -> 3303 | Helped |
Closest papers by meaning
- Learning multiple visual domains with residual adapters (NeurIPS 2017 · similarity 0.89)
- Visual Representation Learning over Latent Domains (ICLR 2022 · similarity 0.88)
- Exploring Sparse Visual Prompt for Domain Adaptive Dense Prediction (AAAI 2024 · similarity 0.86)
- Revisiting Continuity of Image Tokens for Cross-domain Few-shot Learning (ICML 2025 · similarity 0.85)
- START: A Generalized State Space Model with Saliency-Driven Token-Aware Transformation (NeurIPS 2024 · similarity 0.85)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · 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.