Research map / Video / Video segmentation and tracking
Real-World Repetition Estimation by Div, Grad and Curl
Tom F. H. Runia, Cees G. M. Snoek, Arnold W. M. Smeulders
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 |
|---|---|---|---|
| use the proposed flow-signal and wavelet method instead of Pogalin et al. | repetition counting · QUVA Repetition | MAE: 38.5 ± 37.6 -> 23.2 ± 34.4 | Improved |
| use the proposed flow-signal and wavelet method instead of Levy & Wolf | repetition counting · QUVA Repetition | MAE: 48.2 ± 61.5 -> 23.2 ± 34.4 | Improved |
| use the proposed method instead of Levy & Wolf | repetition counting · YTSegments | MAE: 6.5 ± 9.2 -> 10.3 ± 19.8 | Hurt |
| use the proposed method instead of Pogalin et al. | repetition counting · YTSegments | MAE: 21.9 ± 30.1 -> 10.3 ± 19.8 | Improved |
| use time-localized wavelets rather than a global Fourier periodogram | idealized cycle signals from QUVA Repetition · QUVA Repetition | counting performance: null | Improved |
| add diverse flow-based signals with per-video signal selection instead of relying on one fixed signal (ablation) | repetition counting · QUVA Repetition | MAE: 32.9 ± 31.4 -> 10.5 ± 15.7 | Improved |
| use only vertical flow instead of selecting among diverse signals (ablation) | flow-based repetition counting · QUVA Repetition | MAE: 10.5 ± 15.7 -> 32.9 ± 31.4 | Hurt |
| use only horizontal flow instead of selecting among diverse signals (ablation) | flow-based repetition counting · QUVA Repetition | MAE: 10.5 ± 15.7 -> 38.3 ± 31.4 | Hurt |
| use only the divergence signal instead of selecting among diverse signals (ablation) | flow-based repetition counting · QUVA Repetition | MAE: 10.5 ± 15.7 -> 44.9 ± 34.8 | Hurt |
| use only the curl signal instead of selecting among diverse signals (ablation) | flow-based repetition counting · QUVA Repetition | MAE: 10.5 ± 15.7 -> 44.9 ± 34.8 | Hurt |
| use only ∇x Fx instead of selecting among diverse signals (ablation) | flow-based repetition counting · QUVA Repetition | MAE: 10.5 ± 15.7 -> 46.7 ± 30.8 | Hurt |
| use only ∇y Fy instead of selecting among diverse signals (ablation) | flow-based repetition counting · QUVA Repetition | MAE: 10.5 ± 15.7 -> 42.7 ± 39.8 | Hurt |
Closest papers by meaning
- Repetition Estimation (IJCV 2019 · similarity 0.94)
- Dancing with Still Images: Video Distillation via Static-Dynamic Disentanglement (CVPR 2024 · similarity 0.83)
- Emergent Temporal Correspondences from Video Diffusion Transformers (NeurIPS 2025 · similarity 0.82)
- Chirality in Action: Time-Aware Video Representation Learning by Latent Straightening (NeurIPS 2025 · similarity 0.82)
- MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Model (CVPR 2024 · similarity 0.82)
- FlowMo: Variance-Based Flow Guidance for Coherent Motion in Video Generation (NeurIPS 2025 · similarity 0.82)
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.
