Research map / Video / Video segmentation and tracking
Look Before You Match: Instance Understanding Matters in Video Object Segmentation
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 |
|---|---|---|---|
| Add instance understanding and instance-augmented matching to memory-based VOS | ISVOS versus the strongest reported baseline · DAVIS 2017 val | J&F: 86.2 -> 87.1 | Helped |
| Add query enhancement to the query key (ablation) | ISVOS · DAVIS 2017 val | J&F: 85.7 -> 87.1 | Helped |
| Add query enhancement to the query key (ablation) | ISVOS · YouTube-VOS 2018 val | G: 84.4 -> 86.3 | Helped |
| Add multi-path fusion (ablation) | ISVOS · DAVIS 2017 val | J&F: 86.2 -> 87.1 | Helped |
| Add multi-path fusion (ablation) | ISVOS · YouTube-VOS 2018 val | G: 85.6 -> 86.3 | Helped |
| Initialize the instance branch from Mask2Former pretrained on COCO (ablation) | ISVOS · DAVIS 2017 val | J&F: 78.6 -> 82.0 | Helped |
| Jointly train the instance and VOS tasks (ablation) | ISVOS with pretrained Mask2Former weights · DAVIS 2017 val | J&F: 82.0 -> 87.1 | Helped |
| Use multi-scale inference (ablation) | ISVOS · DAVIS 2016 val | J&F: 92.6 -> 92.9 | Helped |
| Use multi-scale inference (ablation) | ISVOS · DAVIS 2017 val | J&F: 87.1 -> 88.6 | Helped |
| Increase training data from DAVIS 2017 only to YouTube-VOS 2019 only (ablation) | ISVOS · DAVIS 2017 val | J&F: 77.5 -> - | Mixed |
| Add static-image pretraining and YouTube-VOS training to DAVIS training (ablation) | ISVOS · DAVIS 2017 val | J&F: 77.5 -> 87.1 | Helped |
| Add BL30K pretraining (ablation) | ISVOS trained on static images, DAVIS, and YouTube-VOS · DAVIS 2017 val | J&F: 87.1 -> 88.2 | Helped |
Closest papers by meaning
- Video Object Segmentation Using Global and Instance Embedding Learning (CVPR 2021 · similarity 0.90)
- Test-time Training for Matching-based Video Object Segmentation (NeurIPS 2023 · similarity 0.90)
- InsPro: Propagating Instance Query and Proposal for Online Video Instance Segmentation (NeurIPS 2022 · similarity 0.90)
- VITA: Video Instance Segmentation via Object Token Association (NeurIPS 2022 · similarity 0.90)
- Point-VOS: Pointing Up Video Object Segmentation (CVPR 2024 · similarity 0.89)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.88)
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.