ActivityNet
A large-scale video benchmark for human activity understanding. Provides samples from 203 activity classes with an average of 137 untrimmed videos per class and 1.41 activity instances per video, for a total of 849 video hours. The benchmark covers a wide range of complex human activities that are of interest to people in their daily living and can be used to compare algorithms for three scenarios: untrimmed video classification, trimmed activity classification, and activity detection.
Progress Over Time
Interactive timeline showing model performance evolution on ActivityNet
State-of-the-art frontier
Open
Proprietary
ActivityNet Leaderboard
1 models
| Context | Cost | License | ||||
|---|---|---|---|---|---|---|
| 1 |  OpenAI | — | 128K | $2.50 / $10.00 |
Notice missing or incorrect data?
FAQ
Common questions about ActivityNet
A large-scale video benchmark for human activity understanding. Provides samples from 203 activity classes with an average of 137 untrimmed videos per class and 1.41 activity instances per video, for a total of 849 video hours. The benchmark covers a wide range of complex human activities that are of interest to people in their daily living and can be used to compare algorithms for three scenarios: untrimmed video classification, trimmed activity classification, and activity detection.
The ActivityNet paper is available at https://openaccess.thecvf.com/content_cvpr_2015/html/Heilbron_ActivityNet_A_Large-Scale_2015_CVPR_paper.html. This paper provides detailed information about the benchmark methodology, dataset creation, and evaluation criteria.
The ActivityNet dataset is available at https://github.com/activitynet/ActivityNet.
The ActivityNet leaderboard ranks 1 AI models based on their performance on this benchmark. Currently, GPT-4o by OpenAI leads with a score of 0.619. The average score across all models is 0.619.
The highest ActivityNet score is 0.619, achieved by GPT-4o from OpenAI.
1 models have been evaluated on the ActivityNet benchmark, with 0 verified results and 1 self-reported results.
ActivityNet is categorized under video and vision. The benchmark evaluates video models.