Research map / Video / Video segmentation and tracking
Real-Time Selfie Video Stabilization
Jiyang Yu, Ravi Ramamoorthi, Keli Cheng, Michel Sarkis, Ning Bi
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 |
|---|---|---|---|
| Use the trained linear network instead of direct optimization of warp-node displacements | Selfie video stabilization · Example videos | Stability: 0.40 -> 0.60 | Improved |
| Add Leaky ReLU activations to the linear network (ablation) | Selfie video stabilization · Example videos, C=32, C=64, and C=128 | Stability: 0.56 -> 0.48; 0.57 -> 0.52; 0.60 -> 0.57 | Hurt |
| Add Tanh activations to the linear network (ablation) | Selfie video stabilization · Example videos, C=32, C=64, and C=128 | Stability: 0.56 -> 0.50; 0.57 -> 0.52; 0.60 -> 0.52 | Hurt |
| Remove foreground mask detection (ablation) | Full selfie stabilization pipeline · Example videos | Stability: 0.60 -> 0.52 | Hurt |
| Increase network filter count from C=32 to C=128, with no activations (ablation) | Selfie video stabilization network · Example videos | Stability: 0.56 -> 0.60 | Improved |
| Use HD or FHD rather than 832 x 448 input (ablation) | Selfie video stabilization · HD (1280x720) and FHD (1920x1080) | Stability: 0.60 -> 0.59; 0.60 -> 0.58 | No clear effect |
| Approximate dense MLS warping with a grid | Selfie video stabilization · Frame warping | warping time per frame: 1s/frame -> 4ms/frame | Improved |
Closest papers by meaning
- Selfie Video Stabilization (ECCV 2018 · similarity 0.93)
- Hybrid Neural Fusion for Full-Frame Video Stabilization (ICCV 2021 · similarity 0.87)
- Minimum Latency Deep Online Video Stabilization (ICCV 2023 · similarity 0.86)
- Robust Video Stabilization by Optimization in CNN Weight Space (CVPR 2019 · similarity 0.86)
- Harnessing Meta-Learning for Improving Full-Frame Video Stabilization (CVPR 2024 · similarity 0.85)
- Instant Video Models: Universal Adapters for Stabilizing Image-Based Networks (NeurIPS 2025 · similarity 0.83)
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.
