Research map / Video / Video segmentation and tracking
Collaborative Spatial-Temporal Modeling for Language-Queried Video Actor Segmentation
Tianrui Hui, Shaofei Huang, Si Liu, Zihan Ding, Guanbin Li, Wenguan Wang, Jizhong Han, Fei Wang
Check 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 |
|---|---|---|---|
| Combine the spatial target-frame encoder with the temporal video encoder (ablation) | single-encoder segmentation model · A2D Sentences | Mean IoU: 47.7/48.2 -> 51.7 | Improved |
| Add LGFS to the combined spatial-temporal encoders (ablation) | spatial-temporal model without LGFS · A2D Sentences | Mean IoU: 51.7 -> 52.9 | Improved |
| Add CMAM to the encoders (ablation) | spatial-temporal model with LGFS · A2D Sentences | Mean IoU: 52.9 -> 56.1 | Improved |
| Use LGFS rather than elementwise addition for spatial-temporal fusion (ablation) | spatial-temporal decoder without CMAM · A2D Sentences | AP 0.5:0.95: 34.5 -> 36.0 | Improved |
| Use LGFS rather than elementwise maximum for spatial-temporal fusion (ablation) | spatial-temporal decoder without CMAM · A2D Sentences | AP 0.5:0.95: 34.8 -> 36.0 | Improved |
| Add CMAM at I4 alongside I5 (ablation) | CMAM inserted at I5 · A2D Sentences | Mean IoU: 53.3 -> 55.0 | Improved |
| Extend CMAM insertion from I5/I4 to include I3, I2, and I1 (ablation) | spatial-temporal encoders with CMAM · A2D Sentences | Mean IoU: 55.0 -> 56.1 | Improved |
| Use ResNet-50 + I3D instead of Inception V3 + I3D (ablation) | collaborative spatial-temporal framework · A2D Sentences | Mean IoU: 56.1 -> 56.4 | Improved |
| Replace I3D with S3D while using Inception V3 (ablation) | collaborative spatial-temporal framework · A2D Sentences | Mean IoU: 56.1 -> 55.0 | Hurt |
| Use the proposed collaborative framework rather than the strongest reported baseline | A2D Sentences segmentation · A2D Sentences | Mean IoU: 53.1 -> 56.1 | Improved |
| Use the proposed collaborative framework rather than CMDy | cross-dataset evaluation without finetuning · J-HMDB Sentences | AP 0.5:0.95: 30.1 -> 33.5 | Improved |
Closest papers by meaning
- Exploiting Temporal State Space Sharing for Video Semantic Segmentation (CVPR 2025 · similarity 0.87)
- Kronecker Mask and Interpretive Prompts are Language-Action Video Learners (ICLR 2025 · similarity 0.87)
- Enhancing Temporal Understanding in Video-LLMs through Stacked Temporal Attention in Vision Encoders (NeurIPS 2025 · similarity 0.87)
- Video-Panda: Parameter-efficient Alignment for Encoder-free Video-Language Models (CVPR 2025 · similarity 0.86)
- The Devil is in Temporal Token: High Quality Video Reasoning Segmentation (CVPR 2025 · similarity 0.86)
- Weakly Supervised Temporal Action Localization via Dual-Prior Collaborative Learning Guided by Multimodal Large Language Models (CVPR 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.
