Chuyển đến nội dung
Bản đồ AI

Hiểu video

Hiểu điều gì diễn ra trong video

VideoTrung cấp #25
đầu vàoVideoVăn bảnVăn bản

Toàn văn của trang này được trình bày bằng tiếng Anh; tiêu đề và phần dẫn đã được bản địa hóa.

NĂNG LỰC NÀY NGHĨA LÀ GÌ

Takes a video, optionally with a question, and outputs a description or answer about content, events and timeline — what happened after three minutes, or whether a movement was performed correctly. It adds a time dimension to image understanding and must distinguish events with similar frames but different order. Input is usually sampled into frames and compressed before entering the model.

Làm ra sao về mặt kỹ thuật

Frames are sampled on a fixed or adaptive schedule, a vision encoder turns each into tokens, and frame order with timestamps enters a multimodal Transformer whose language side produces text. Long videos are managed with hierarchical summarisation, keyframe selection or memory mechanisms to cap the token budget. Precise localisation is handled by a dedicated temporal grounding head or by having the model output a time interval.

Sản phẩm tiêu biểu

5

Tổ chức liên quan

Cách dùng tiêu biểu

  • Event search in surveillance and security footage
  • Technique review in sport and fitness
  • Overviews of lectures and meetings
  • Moderation and locating offending segments

Đánh giá nó tốt hay không thế nào

Video QA accuracy
Share answered correctly on annotated questions
Temporal grounding mIoU
Mean IoU between predicted and true time intervals
Video captioning CIDEr
Semantic match of captions against references

Ranh giới và điểm khó

  • Frame sampling discards mid-video detail, and one-off events are the easiest to miss
  • Fine temporal ordering — who did what first — is vague and often reversed
  • Judgement is unstable when audio and picture conflict, as dubbed audio can make the model waver

Các khái niệm đằng sau