Akshita Gupta

A new model enhances action recognition in untrimmed videos while minimizing memory use.

A new method improves video action recognition using contextual language.

A new method generates speech from videos, enhancing dubbing and language learning.