A Short Note on Evaluating RepNet for Temporal Repetition Counting in Videos

Debidatta Dwibedi, Yusuf Aytar, Jonathan Tompson, Pierre Sermanet, Andrew Zisserman

2024-11-13Repetitive Action Counting

Abstract

We discuss some consistent issues on how RepNet has been evaluated in various papers. As a way to mitigate these issues, we report RepNet performance results on different datasets, and release evaluation code and the RepNet checkpoint to obtain these results. Code URL: https://github.com/google-research/google-research/blob/master/repnet/

Results

Task	Dataset	Metric	Value	Model
Repetitive Action Counting	Countix	MAE	0.3002	RepNet
Repetitive Action Counting	Countix	OBO	0.7047	RepNet
Repetitive Action Counting	RepCount	MAE	0.3308	RepNet
Repetitive Action Counting	RepCount	OBO	0.5329	RepNet
Repetitive Action Counting	UCFRep	MAE	0.2088	RepNet
Repetitive Action Counting	UCFRep	OBO	0.7333	RepNet

Related Papers

CountLLM: Towards Generalizable Repetitive Action Counting via Large Language Model2025-03-22 Localization-Aware Multi-Scale Representation Learning for Repetitive Action Counting2025-01-13 Repetitive Action Counting with Hybrid Temporal Relation Modeling2024-12-10 MultiCounter: Multiple Action Agnostic Repetition Counting in Untrimmed Videos2024-09-06 GMFL-Net: A Global Multi-geometric Feature Learning Network for Repetitive Action Counting2024-08-31 Rethinking temporal self-similarity for repetitive action counting2024-07-12 FCA-RAC: First Cycle Annotated Repetitive Action Counting2024-06-18 Skim then Focus: Integrating Contextual and Fine-grained Views for Repetitive Action Counting2024-06-13