Research map / Video / Video segmentation and tracking
PRISM: Video Dataset Condensation with Progressive Refinement and Insertion for Sparse Motion
Jaehyun Choi, Jiwan Hur, Gyojin Han, Jaemyung Yu, Junmo Kim
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 |
|---|---|---|---|
| PRISM vs prior video condensation method (Wang et al.) | PRISM · miniUCF VPC 1 | top-1 accuracy: 17.5 -> 17.9 | Improved |
| PRISM vs Wang et al., storage | PRISM · miniUCF VPC 1 | storage (MB): 94 -> 20 | Improved |
| PRISM vs image-based DM at low-data regime | PRISM · Kinetics-400 VPC 5 | top-5 accuracy: 9.1 -> 8.1 | Hurt |
| PRISM vs DM | PRISM · Kinetics-400 VPC 1 | top-5 accuracy: 6.3 -> 7.1 | Improved |
| PRISM vs Herding at VPC 10 | PRISM · miniUCF VPC 10 | top-1 accuracy: 33.7 -> 31.0 | Hurt |
| PRISM vs Wang et al. on other architectures | PRISM · miniUCF VPC 1 CNN+GRU | top-1 accuracy: 12.0 -> 18.9 | Improved |
| start with more initial key-frames (8 vs 2) (ablation) | PRISM · miniUCF | top-1 accuracy: 17.9 -> 15.3 | Hurt |
| add adaptive frame insertion (ablation) | PRISM · miniUCF | top-1 accuracy: 15.8 -> 17.9 | Improved |
| add adaptive frame insertion (ablation) | PRISM · HMDB51 | top-1 accuracy: 6.1 -> 7.5 | Improved |
| choose frames by negative gradient instead of random (ablation) | PRISM · miniUCF | top-1 accuracy: 16.8 -> 17.9 | Improved |
| use cosine similarity instead of L2 distance (ablation) | PRISM · HMDB51 | top-1 accuracy: 6.0 -> 7.5 | Improved |
| add warm-up phase (ablation) | PRISM · miniUCF | top-1 accuracy: 16.1 -> 17.9 | Improved |
Closest papers by meaning
- 4D3R: Motion-Aware Neural Reconstruction and Rendering of Dynamic Scenes from Monocular Videos (NeurIPS 2025 · similarity 0.86)
- Dancing with Still Images: Video Distillation via Static-Dynamic Disentanglement (CVPR 2024 · similarity 0.85)
- HiCoM: Hierarchical Coherent Motion for Dynamic Streamable Scenes with 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.85)
- World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video (NeurIPS 2026 · similarity 0.85)
- Instant4D: 4D Gaussian Splatting in Minutes (NeurIPS 2025 · similarity 0.85)
- TTVidT: Decoupling the Temporal Axis for Efficient Motion-Centric Video Pretraining (NeurIPS 2026 · similarity 0.85)
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.
