Research map / Video / Video segmentation and tracking
SMC++: Masked Learning of Unsupervised Video Semantic Compression
Yuan Tian, Xiaoyue Ling, Cong Geng, Qiang Hu, Guo Lu, Guangtao Zhai
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 masked learning instead of MoCoV3 contrastive objective (ablation) | SMC · UCF101-TSM @0.02bpp | Top1 accuracy: 75.20% -> 76.48% | Improved |
| add masked learning instead of MoCoV3 contrastive objective (ablation) | SMC · MOT17-ByteTrack @0.01bpp | MOTA: 67.62% -> 70.84% | Improved |
| add non-semantic suppression (NSS) loss (ablation) | SMC · UCF101-TSM | Top1 accuracy: 69.57% -> 72.96% | Improved |
| add semantic feature as GMM condition (ablation) | SMC · UCF101-TSM | Top1 accuracy: 71.38 -> 72.96 | Improved |
| use masking ratio 90% instead of 50% (ablation) | SMC · UCF101-TSM @0.02bpp | Top1 accuracy: 73.85 -> 76.48 | Improved |
| use masking ratio 90% instead of 95% (ablation) | SMC · UCF101-TSM @0.02bpp | Top1 accuracy: 76.08 -> 76.48 | No clear effect |
| predict DINOv2 features instead of raw pixels (ablation) | SMC · UCF101-TSM | Top1 accuracy: 76.48 -> 77.00 at 0.02bpp; 89.11 -> 89.51 at 0.06bpp | Mixed |
| add semantic compensation operation (ablation) | SMC · UCF101-TSM @0.02bpp | Top1 accuracy: 16% drop when removed | Improved |
| add LPIPS loss term (ablation) | SMC · UCF101-TSM @0.02bpp | Top1 accuracy: about 15% drop when removed | Improved |
| add GAN loss term (ablation) | SMC · UCF101-TSM @0.02bpp | Top1 accuracy: about 5% drop when removed | Improved |
| add masked motion objective with RGB difference target (ablation) | SMC-adv · VOS DAVIS2017 @0.02bpp | J&F: 79.30 -> 79.76 | Improved |
| add Blue-Tr module (ablation) | SMC-adv+Motion · VOS DAVIS2017 @0.02bpp | J&F: 79.76 -> 81.06 | Improved |
Closest papers by meaning
- One Token per Highly Selective Frame: Towards Extreme Compression for Long Video Understanding (NeurIPS 2025 · similarity 0.87)
- Towards Unified Human Perception and Machine Understanding: Token Flow Guided Compression Framework (CVPR 2026 · similarity 0.87)
- Video Language Model Pretraining with Spatio-temporal Masking (CVPR 2025 · similarity 0.86)
- Extending Video Masked Autoencoders to 128 frames (NeurIPS 2024 · similarity 0.86)
- SMILE: Infusing Spatial and Motion Semantics in Masked Video Learning (CVPR 2025 · similarity 0.86)
- EVEREST: Efficient Masked Video Autoencoder by Removing Redundant Spatiotemporal Tokens (ICML 2024 · 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.
