Listen first―look only when it matters (IMAGE)

Japan Advanced Institute of Science and Technology

Listen first―look only when it matters

Caption

The AI first analyzes the audio to identify a few-second intervals worth examining. Visual analysis is conducted only for the selected intervals, while the remaining frames are ignored. The extracted visual information is then combined with the audio and transcript data to generate the final evaluation.

Credit

Professor Shogo Okada from JAIST, Japan

Usage Restrictions

Cannot be reused without permission

License

Original content

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.