图像识别模型的训练及图像识别方法、装置、介质、设备
By preprocessing and enhancing the image recognition model, and training the network model with the target loss function, a multi-dimensional image recognition model is constructed, which solves the problem of low recognition rate of handwritten formulas and improves the accuracy of the model.
CN116434252BActive Publication Date: 2026-07-17BOE TECHNOLOGY GROUP CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BOE TECHNOLOGY GROUP CO LTD
- Filing Date
- 2023-04-10
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing image recognition models have low recognition rates and low accuracy for handwritten formulas.
Method used
By acquiring and preprocessing original historical images, image recognition and enhancement are performed using the network model to be trained. A target loss function is constructed, and the network model is trained based on this function to build a multi-dimensional image recognition model. The model is then trained in conjunction with the image enhancement results.
Benefits of technology
This improves the accuracy of trained network models and the precision of image recognition models, solving the problem of low model accuracy in existing technologies.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116434252B_ABST
Abstract
本公开是关于一种图像识别模型的训练及图像识别方法、装置、介质、设备，涉及计算机视觉技术领域，该图像识别模型的训练方法包括：获取原始历史图像，并对原始历史图像进行预处理，得到目标历史图像；基于待训练的网络模型对所述目标历史图像进行图像识别以及图像增强处理，得到图第一公式预测结果以及图像增强结果；基于所述第一公式预测结果以及图像增强结果构建目标损失函数；基于目标损失函数对所述待训练的网络模型进行训练，得到训练完成的网络模型，并基于训练完成的网络模型，构建多维度的图像识别模型。本公开提高了图像识别模型的精确度。
Need to check novelty before this filing date? Find Prior Art