You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A clean, modular implementation of Moment-DETR for language-based temporal video grounding on custom datasets. Given a video and a natural-language query, the model predicts the start/end timestamps of the described moment, with a streamlined pipeline for preprocessing, feature extraction, training, and evaluation.