Research map / Detection and segmentation / Open-set and few-shot detection
Zoom-Net: Mining Deep Feature Interactions for Visual Relationship Recognition
Guojun Yin, Lu Sheng, Bin Liu, Nenghai Yu, Xiaogang Wang, Jing Shao, Chen Change Loy
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 message passing between branches (CA-M over A-M) (ablation) | Zoom-Net · Visual Genome | predicate Acc@1: 35.89 -> 46.81 | Improved |
| add spatiality-aware Contrastive and Pyramid ROI pooling (SCA-M over CA-M) (ablation) | Zoom-Net · Visual Genome | relationship Acc@1: 9.46 -> 10.51 | Improved |
| stack two SCA-Ms instead of one (ablation) | Zoom-Net · Visual Genome | relationship Acc@1: 10.51 -> 11.42 | Improved |
| stack two SCA-Ms instead of one (ablation) | Zoom-Net · Visual Genome | relationship Rec@100: 23.51 -> 25.07 | Improved |
| add both IH-tree levels H(1) and H(2) (ablation) | Zoom-Net · Visual Genome | relationship Acc@1: 9.92 -> 11.42 | Improved |
| add second IH-tree level H(2) (semantic clustering) (ablation) | Zoom-Net · Visual Genome | relationship Acc@1: 10.76 -> 11.42 | Improved |
| add IH-tree to subject classification (ablation) | Zoom-Net · Visual Genome | subject Acc@1: 36.52 -> 38.94 | Improved |
| replace ViP components with SCA-M | ViP · Visual Genome | relationship Acc@1: 8.16 -> 10.65 | Improved |
| replace ViP label structure with IH-tree | ViP · Visual Genome | relationship Acc@1: 8.16 -> 9.97 | Improved |
| add both SCA-M and IH-tree to ViP | ViP · Visual Genome | relationship Acc@1: 8.16 -> 11.79 | Improved |
| replace visual features in CAI with SCA-M | CAI · VRD (k=1) | predicate Rec@50: 53.59 -> 55.98 | Improved |
| replace visual features in CAI with SCA-M | CAI · VRD (k=1) | relationship Rec@50: 15.63 -> 19.54 | Improved |
Later papers that compared with this method
6 later accepted papers report ZoomNet as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- Referring Camouflaged Object Detection (TPAMI 2025 · better on 1 of 1 shared result)
- Feature Shrinkage Pyramid for Camouflaged Object Detection With Transformers (CVPR 2023 · better on 6 of 6 shared results)
- Stereo Neural Vernier Caliper (AAAI 2022 · better on 1 of 1 shared result)
- RTS3D: Real-time Stereo 3D Detection from 4D Feature-Consistency Embedding Space for Autonomous Driving (AAAI 2021 · better on 0 of 2 shared results)
- Visual Relationship Detection Using Part-and-Sum Transformers With Composite Queries (ICCV 2021 · better on 1 of 1 shared result)
- Localize, Assemble, and Predicate: Contextual Object Proposal Embedding for Visual Relation Detection (AAAI 2020 · better on 7 of 8 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PPRFCN — PPR-FCN: Weakly Supervised Visual Relation Detection via Parallel Pairwise R-FCN (ICCV 2017) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Large-Scale Visual Relationship Understanding (AAAI 2019 · similarity 0.91)
- Visual Relationship Detection With Internal and External Linguistic Knowledge Distillation (ICCV 2017 · similarity 0.91)
- Visual Relationship Detection with Language Priors (ECCV 2016 · similarity 0.90)
- Localize, Assemble, and Predicate: Contextual Object Proposal Embedding for Visual Relation Detection (AAAI 2020 · similarity 0.90)
- Scene-Graph ViT: End-to-End Open-Vocabulary Visual Relationship Detection (ECCV 2024 · similarity 0.90)
- Explanation-based Weakly-supervised Learning of Visual Relations with Graph Networks (ECCV 2020 · similarity 0.89)
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.
