Research map / Video / Video segmentation and tracking
Scalable Neural Video Representations with Learnable Positional Features
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use learnable 2D keyframes and sparse 3D positional features | NVP · UVG-HD, ~5 minutes | PSNR: 33.15 ± 3.19 -> 34.57 ± 2.62 | Helped |
| Use learnable 2D keyframes and sparse 3D positional features | NVP · UVG-HD, ~5 minutes | LPIPS: 0.226 ± 0.112 -> 0.190 ± 0.100 | Helped |
| Use NVP instead of NeRV for frame interpolation | Video representations on Big Buck Bunny · Big Buck Bunny, frame interpolation | PSNR: 23.05 -> 33.76 | Helped |
| Add learnable keyframes (ablation) | NVP · UVG-HD, 150K iterations | PSNR: 31.21 ± 2.80 -> 38.89 ± 2.11 | Helped |
| Add sparse positional features (ablation) | NVP · UVG-HD, 150K iterations | PSNR: 32.44 ± 4.48 -> 38.89 ± 2.11 | Helped |
| Add modulation (ablation) | NVP · UVG-HD, 1.5K iterations | PSNR: 32.15 ± 3.08 -> 34.85 ± 2.69 | Helped |
| Add modulation (ablation) | NVP · UVG-HD, 150K iterations | PSNR: 38.04 ± 2.27 -> 38.89 ± 2.11 | Helped |
| Include keyframes across spatial directions in addition to temporal keyframes (ablation) | NVP · Jockey | PSNR | Helped |
| Concatenate nearby sparse positional feature codes (ablation) | NVP · ReadySetGo | reconstruction quality | Helped |
| Upsample sparse positional features (ablation) | NVP · unseen video coordinates | smoothness | Helped |
| Upsample sparse positional features (ablation) | NVP · NVP training, per iteration | time: 0.291s -> 0.469s | Hurt |
| Use NVP-M rather than NeRV-M | Video representations on Big Buck Bunny · Big Buck Bunny | PSNR: 36.50 -> 36.49 | No clear effect |
Closest papers by meaning
- Neural Video Compression with Feature Modulation (CVPR 2024 · similarity 0.88)
- CoANeRV: Coordinate-Aware Token-Space Neural Video Representation (NeurIPS 2026 · similarity 0.88)
- Combining Frame and GOP Embeddings for Neural Video Representation (CVPR 2024 · similarity 0.88)
- Towards Practical Real-Time Neural Video Compression (CVPR 2025 · similarity 0.87)
- NVRC: Neural Video Representation Compression (NeurIPS 2024 · similarity 0.87)
- PNeRV: Enhancing Spatial Consistency via Pyramidal Neural Representation for Videos (CVPR 2024 · similarity 0.87)
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.