Research map / Video / Video segmentation and tracking
Learning Discriminative Feature with CRF for Unsupervised Video Object Segmentation
Mingmin Zhen, Shiwei Li, Lei Zhou, Jiaxiang Shang, Haoan Feng, Tian Fang, Long Quan
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 DFM and ATM (discriminative features with attention) (ablation) | DeepLabv3 baseline · DAVIS16 | J mean: 76.7 -> 79.5 | Improved |
| add CRF refinement of attention map (ablation) | DFNet · DAVIS16 | J mean: 79.5 -> 80.4 | Improved |
| add multiple-scale inference (ablation) | DFNet · DAVIS16 | J mean: 80.4 -> 81.1 | Improved |
| add instance pruning post-processing (ablation) | DFNet · DAVIS16 | J mean: 81.1 -> 83.4 | Improved |
| increase number of D-features from 128 to 2048 (ablation) | DFNet · DAVIS16 | J mean: 78.3 -> 80.4 | Improved |
| increase number of D-features from 128 to 256 (ablation) | DFNet · DAVIS16 | J mean: 78.3 -> 79.0 | Improved |
| increase input images at inference from 1 to 4 (ablation) | DFNet · DAVIS16 | J mean: 79.4 -> 80.4 | Improved |
| increase input images from 4 to 10 (ablation) | DFNet · DAVIS16 | J mean: 80.4 -> 80.4 | No clear effect |
| increase input images at inference from 1 to 8 (ablation) | DFNet · PASCAL VOC co-segmentation | Avg. J (%): 61.4 -> 65.4 | Improved |
| increase input images from 8 to 10 (ablation) | DFNet · PASCAL VOC co-segmentation | Avg. J (%): 65.4 -> 65.4 | No clear effect |
| use DFNet instead of AnDiff | AnDiff · DAVIS16 test | J mean: 81.7 -> 83.4 | Improved |
| use DFNet instead of Li19 for co-segmentation | Li19 · PASCAL VOC co-segmentation | Jaccard index: 0.63 -> 0.69 | Improved |
Closest papers by meaning
- Unsupervised Object Segmentation in Video by Efficient Selection of Highly Probable Positive Features (ICCV 2017 · similarity 0.88)
- A Survey on Deep Learning Technique for Video Segmentation (TPAMI 2022 · similarity 0.87)
- Video K-Net: A Simple, Strong, and Unified Baseline for Video Segmentation (CVPR 2022 · similarity 0.87)
- DyStaB: Unsupervised Object Segmentation via Dynamic-Static Bootstrapping (CVPR 2021 · similarity 0.87)
- From ViT Features to Training-free Video Object Segmentation via Streaming-data Mixture Models (NeurIPS 2023 · similarity 0.87)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.86)
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.
