Research map / Detection and segmentation / CNN features
Tasks Integrated Networks: Joint Detection and Retrieval for Image Search
Lei Zhang, Zhenwei He, Yi Yang, Liang Wang, Xinbo Gao
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 C2HEP loss in place of HEP (ablation) | I-Net · CUHK-SYSU | mAP (%): 79.5 -> 80.9 | Improved |
| add C2HEP loss in place of HEP (ablation) | DC-I-Net · CUHK-SYSU | mAP (%): 83.7 -> 81.0 when removed (mAP degraded by 2.7%) | Improved |
| add C2HEP loss in place of HEP (ablation) | DC-I-Net · CUHK-SYSU | Top-1 (%): 85.8 -> 83.0 when removed | Improved |
| use OLP loss instead of triplet loss (with HEP) (ablation) | DC-I-Net · CUHK-SYSU | mAP (%): 67.8 -> 81.9 | Improved |
| add C2HEP to OLP-only training (ablation) | DC-I-Net · CUHK-SYSU | mAP (%): 81.3 -> 83.7 | Improved |
| add OLP to C2HEP-only training (ablation) | DC-I-Net · CUHK-SYSU | mAP (%): 82.2 -> 83.7 | Improved |
| add HEP to OLP-only training (ablation) | DC-I-Net · CUHK-SYSU | mAP (%): 81.3 -> 81.9 | Improved |
| DC architecture (separate detection and re-id layers, refined boxes) instead of I-Net (ablation) | I-Net (HEP, VGG16) · CUHK-SYSU | mAP (%): 79.5 -> 81.0 | Improved |
| replace VGG16 backbone with ResNet50 | DC-I-Net · CUHK-SYSU | mAP (%): 83.7 -> 86.2 | Improved |
| add detection module before search (ablation) | DC-I-Net · WebTattoo | mAP (%): 30.4 -> 48.5 | Improved |
| add detection module before search (ablation) | I-Net · WebTattoo | mAP (%): 23.5 -> 41.7 | Improved |
| add detection module before search (ablation) | OIM · WebTattoo | mAP (%): 21.5 -> 38.2 | Improved |
Closest papers by meaning
- A Modulation Module for Multi-task Learning with Applications in Image Retrieval (ECCV 2018 · similarity 0.87)
- Hierarchical Online Instance Matching for Person Search (AAAI 2020 · similarity 0.86)
- End-to-End Learning of Deep Visual Representations for Image Retrieval (IJCV 2017 · similarity 0.85)
- Divide and Conquer: Hybrid Pre-training for Person Search (AAAI 2024 · similarity 0.85)
- Instance-Level Composed Image Retrieval (NeurIPS 2025 · similarity 0.85)
- MatchDet: A Collaborative Framework for Image Matching and Object Detection (AAAI 2024 · similarity 0.84)
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.
