Object-Centric Video Feature Feature Extraction toward eSports Video Captioning

Tsunehiko TANAKA

We propose a new Tubelet Action Detection model utilizing object queries.