Spatio-temporal video grounding is a computer vision and natural language processing (NLP) task that involves linking textual descriptions to specific spatio-temporal regions or moments in a video. In other words, it aims to determine which parts of a video correspond to a given textual query or description. This task is essential for various applications, including video summarization, content-based video retrieval, video captioning, and more.
데이터셋
출처에 같은 태스크 이름으로 연결된 데이터셋입니다. 동명 태스크가 포함될 수 있으므로, 비교할 때는 평가 결과의 조건도 확인하세요.
The VidSTG dataset is a spatio-temporal video grounding dataset constructed based on the video relation dataset VidOR. VidOR contains 7,000, 835 and 2,165 videos for training, validation and testing, respectively. The goal of the Spatio-Temporal Video Grounding task (STVG) is to localize the spatio-temporal section of an untrimmed video that matches a given sentence depicting an object. VidSTG contains 5,563, 618, and 743 videos for training, validation, and testing, respectively. Source: https://github.com/Guaranteer/VidSTG-Dataset Image Source: https://github.com/Guaranteer/VidSTG-Dataset
We have added data and cleaned the labels in HC-STVG to build the HC-STVG2.0. While the original database contained 5660 videos, the new database has been re-annotated and modified and now contains 16,000 + videos for this challenge.
The newly proposed HC-STVG task aims to localize the target person spatio-temporally in an untrimmed video. For this task, we collect a new benchmark dataset, which has spatio temporal annotations related to the target persons in complex multi-person scenes, together with full interaction and rich action information.