Research map / Video / Video segmentation and tracking
Zero-Shot Video Object Segmentation via Attentive Graph Neural Networks
Wenguan Wang, Xiankai Lu, Jianbing Shen, David J. Crandall, Ling Shao
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 the attentive graph neural network over the DeepLabV3 backbone (ablation) | DeepLabV3 · DAVIS16 val | mean J: 72.2 -> 80.7 | Improved |
| add gated message aggregation (ablation) | AGNN · DAVIS16 val | mean J: 80.1 -> 80.7 | Improved |
| increase message passing iterations from 1 to 3 (ablation) | AGNN · DAVIS16 val | mean J: 78.7 -> 80.7 | Improved |
| increase message passing iterations from 2 to 3 (ablation) | AGNN · DAVIS16 val | mean J: 79.1 -> 80.7 | Improved |
| increase message passing iterations from 3 to 4 (ablation) | AGNN · DAVIS16 val | mean J: 80.7 -> 80.7 | No clear effect |
| increase inference nodes N' from 3 to 5 (ablation) | AGNN · DAVIS16 val | mean J: 79.6 -> 80.7 | Improved |
| increase inference nodes N' from 5 to 7 (ablation) | AGNN · DAVIS16 val | mean J: 80.7 -> 80.7 | No clear effect |
| add CRF post-processing (ablation) | AGNN · DAVIS16 val | mean J: 78.9 -> 80.7 | Improved |
| use AGNN instead of PDB with same training protocol (ablation) | PDB · DAVIS16 val | mean J: 77.2 -> 80.7 | Improved |
| model multi-image high-order relations instead of per-image or pairwise (FCNs vs AGNN) (ablation) | FCNs · PASCAL VOC co-segmentation | mean J: 55.21 -> 60.78 | Improved |
Closest papers by meaning
- Segmenting Objects From Relational Visual Data (TPAMI 2021 · similarity 0.93)
- Video Object Segmentation with Episodic Graph Memory Networks (ECCV 2020 · similarity 0.88)
- Motion-Attentive Transition for Zero-Shot Video Object Segmentation (AAAI 2020 · similarity 0.88)
- Zero-Shot Video Object Segmentation with Co-Attention Siamese Networks (TPAMI 2020 · similarity 0.86)
- One-Shot Video Object Segmentation (CVPR 2017 · similarity 0.86)
- Bridge the Points: Graph-based Few-shot Segment Anything Semantically (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.
