基于大语言模型的交互式视频检索方法和系统
By introducing a large language model into interactive video retrieval and generating questions based on user feedback, a deeper understanding of user intent is achieved, solving the problem of long retrieval time in existing technologies and realizing more efficient and accurate video retrieval results.
CN118394969BActive Publication Date: 2026-07-17WUHAN UNIV
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUHAN UNIV
- Filing Date
- 2024-01-09
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing interactive video retrieval technologies struggle to accurately understand users' search intent, resulting in time-consuming iterative retrieval processes and an inability to quickly obtain results that meet user needs.
Method used
By introducing a large language model, extracting video features and combining them with user feedback to generate questions, we can gain a deeper understanding of user intent and update query terms to improve retrieval accuracy and efficiency.
Benefits of technology
By using a large language model to assist interactive retrieval, the accuracy and efficiency of video retrieval are improved, better meeting the specific needs of users.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN118394969B_ABST
Abstract
本发明提供一种基于大语言模型的交互式的视频检索方法和系统。本发明利用预训练的大语言模型作为检索的辅助工具，生成相关问题来询问用户的检索意图，并根据用户的反馈情况来实时更新查询内容，通过与用户的交互，进一步细化查询并提供更准确的结果。本发明首先对对数据集进行分帧和提取特征，针对不同的任务类型，对用户的查询条件进行不同的处理，同时进行了相似度分数计算和结果排序。用户可以对检索结果进行反馈包括标记正负样本、添加到提交列表等操作。之后大语言模型可以在用户的提示词的引导下来执行生成问题和更新查询内容的任务，提高检索结果的准确率和检索的高效性。并且这一过程是可以循环进行的，直至用户查找到自己就满意的结果。
Need to check novelty before this filing date? Find Prior Art