Research map / Graph learning / Knowledge and temporal graphs
Fashion Retrieval via Graph Reasoning Networks on a Similarity Pyramid
Zhanghui Kuang, Yiming Gao, Guanbin Li, Ping Luo, Yimin Chen, Liang Lin, Wayne Zhang
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 graph reasoning over the similarity pyramid (ablation) | GRNet · DeepFashion | top-1 accuracy: 14.06 -> 25.73 | Improved |
| add inter-scale connections (ablation) | GRNet · DeepFashion | top-1 accuracy: 24.58 -> 25.73 | Improved |
| add intra-scale connections (ablation) | GRNet · DeepFashion | top-1 accuracy: 24.79 -> 25.73 | Improved |
| add local similarity scales 1x2, 2x1 and 2x2 (ablation) | GRNet · DeepFashion | top-1 accuracy: 14.06 -> 22.60 | Improved |
| add local similarity scales 3x1, 1x3 and 3x3 to the smaller scales (ablation) | GRNet · DeepFashion | top-1 accuracy: 22.60 -> 25.73 | Improved |
| increase graph convolution layers from 1 to 3 (ablation) | GRNet · DeepFashion | top-1 accuracy: 16.8 -> 22.8 -> 25.7 | Improved |
| increase graph convolution layers beyond 3 (ablation) | GRNet · DeepFashion | top-1 accuracy | Hurt |
| reduce projection dimension D from 512 to 128 (ablation) | GRNet · DeepFashion | top-1 accuracy: 25.73 -> 24.69 | Hurt |
| increase channel number C' from 128 to 512 (ablation) | GRNet · DeepFashion | top-1 accuracy: 25.73 -> 25.92 | No clear effect |
| use GRNet instead of KPM | KPM · DeepFashion | top-1 accuracy: 21.3 -> 25.7 | Improved |
| use GRNet instead of KPM | KPM · FindFashion Hard-View | top-20 accuracy: 45.8 -> 57.9 | Improved |
| use GRNet instead of DREML | DREML(192,48) · FindFashion Hard-Occlusion | top-20 accuracy: 31.3 -> 35.0 | Improved |
Closest papers by meaning
- DeepFashion: Powering Robust Clothes Recognition and Retrieval With Rich Annotations (CVPR 2016 · similarity 0.88)
- Learning Attribute-Driven Disentangled Representations for Interactive Fashion Retrieval (ICCV 2021 · similarity 0.88)
- Which Is Plagiarism: Fashion Image Retrieval Based on Regional Representation for Design Protection (CVPR 2020 · similarity 0.87)
- DeepFashion2: A Versatile Benchmark for Detection, Pose Estimation, Segmentation and Re-Identification of Clothing Images (CVPR 2019 · similarity 0.87)
- Learning to Match on Graph for Fashion Compatibility Modeling (AAAI 2020 · similarity 0.87)
- FashionERN: Enhance-and-Refine Network for Composed Fashion Image Retrieval (AAAI 2024 · similarity 0.87)
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.
