14 datasets found

MAD: A large-scale benchmark for long-form video temporal grounding, containing over 384K natural language queries that derived from high-quality audio description of mainstream...

The TACoS Speech dataset contains a large amount of open-world videos with more shot transitions.

The Charades-STA Speech dataset contains a large amount of open-world videos with more shot transitions.

The ActivityNet Speech dataset contains a large amount of open-world videos with more shot transitions.

A dataset of videos with multiple sentence descriptions, used for activity recognition and video description tasks.

Dense regression network for video grounding

Semantic conditioned dynamic modulation for temporal sentence grounding in videos

Multilevel language and vision integration for text-to-clip retrieval

Tall: Temporal activity localization via language query.

Support-Set Based Cross-Supervision for Video Grounding

Charades-STA dataset contains 12,408/3720 segment-sentence pairs and 5338/1334 videos in training and test set, respectively.

Given an untrimmed video and a language query, video grounding aims to localize the time interval by understanding the text and video simultaneously.

Localizing moments in video with natural language

The ActivityNet Captions is a benchmark dataset proposed for dense video captioning. There are 20K untrimmed videos in total, and each video has several annotated segments with...