信息提取方法、装置、设备及存储介质
By combining a pre-set recognition model and an extraction template, the problem of low information extraction efficiency caused by the large computational load of deep learning models is solved, achieving faster information extraction speed and higher efficiency.
CN117292383BActive Publication Date: 2026-07-17CHINA MERCHANTS BANK
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA MERCHANTS BANK
- Filing Date
- 2023-09-21
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In existing technologies, when using deep learning models to identify and extract key information from user-input images, the large amount of computation results in low overall information extraction efficiency.
Method used
The system uses a preset recognition model to recognize user-uploaded image data and extracts key information based on search information and matching methods using a preset extraction template. This reduces reliance on deep learning models and uses INT8 precision instead of FP32 precision to reduce computing power requirements and speed up computation.
Benefits of technology
It improves the efficiency of information extraction, reduces the amount of calculation and workload, and increases the overall speed of information extraction.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN117292383B_ABST
Abstract
本申请公开了一种信息提取方法、装置、设备及存储介质，该方法包括步骤：获取用户上传的第一图像数据；将所述第一图像数据输入至预设识别模型，基于所述预设识别模型，对所述第一图像数据进行识别，得到第一识别数据；通过预设提取模板提取所述第一识别数据中的目标信息，其中，所述预设提取模板是基于搜索信息并进行匹配的方式来提取目标信息的。本申请提高了对所接收图像数据的信息提取效率。
Need to check novelty before this filing date? Find Prior Art