Research map / Video / Video segmentation and tracking
BiVM: Accurate Binarized Neural Network for Efficient Video Matting
Haotong Qin, Xianglong Liu, Xudong Ma, Lei Ke, Yulun Zhang, Jie Luo, Michele Magno
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 EBB encoder (evolvable binarized block) (ablation) | 1-bit RVM · VM ablation (low-res) | MAD: 28.49 -> 13.46 | Improved |
| add LBM mimicking optimization (ablation) | 1-bit RVM · VM ablation (low-res) | MAD: 28.49 -> 24.65 | Improved |
| add SHB decoder alone (ablation) | 1-bit RVM · VM ablation (low-res) | MAD: 28.49 -> 191.89 | Hurt |
| add SHB decoder alone (efficiency) (ablation) | 1-bit RVM · VM ablation (low-res) | FLOPs (G): 0.55 -> 0.27 | Improved |
| add EBB, SHB and LBM together (ablation) | 1-bit RVM · VM ablation (low-res) | MAD: 28.49 -> 11.82 | Improved |
| add SHB and LBM to EBB model (FLOPs) (ablation) | 1-bit RVM with EBB · VM ablation (low-res) | FLOPs (G): 0.57 -> 0.32 | Improved |
| add EBB to 32-bit RVM (ablation) | 32-bit RVM · VM ablation (low-res) | MAD: 6.08 -> 5.91 | Improved |
| use BiVM instead of RVM-ReActNet | 1-bit video matting · D646 (low-res comparison) | MAD: 56.41 -> 31.12 | Improved |
| use BiVM instead of BiMatting | 1-bit video matting · VM (low-res comparison) | MAD: 12.82 -> 11.82 | Improved |
| binarize RVM with BNN or ReCU directly | RVM · VM (low-res comparison) | MAD: 6.08 -> 189.13 | Hurt |
| binarize to BiVM 1-bit (latency) | RVM 32-bit · HUAWEI P40 Pro, 512x512 | latency (ms): 1313.55 -> 102.98 | Improved |
Closest papers by meaning
- BiMatting: Efficient Video Matting via Binarization (NeurIPS 2023 · similarity 0.93)
- Deep Video Matting via Spatio-Temporal Alignment and Aggregation (CVPR 2021 · similarity 0.87)
- MatAnyone: Stable Video Matting with Consistent Memory Propagation (CVPR 2025 · similarity 0.85)
- BiMaCoSR: Binary One-Step Diffusion Model Leveraging Flexible Matrix Compression for Real Super-Resolution (ICML 2025 · similarity 0.84)
- Revisiting Context Aggregation for Image Matting (ICML 2024 · similarity 0.84)
- Binarized Low-light Raw Video Enhancement (CVPR 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.
