基于深度学习的手写数字图像识别模型研究及其浏览器服务平台搭建

Research on Handwritten Digital Image Recognition Model Based on Deep Learning and Its Browser Service Platform Construction

为制作手写数字图像数据集,使用OpenCV图像识别技术来分割表格并提取网格图片。对收集的100份手写数字表格,进行表格识别与去除,再对去除表格后的图像进行分割,得到10000张28×28的手写数字图像,作为数据集。用LeNet-5模型网络检测数据集的数据质量优劣情况,得出自制数据集的测试集得分为98.30%,并从混淆矩阵与分类报告输出结果中发现,数据集是可用的。使用HTML构建网站,Flask框架来实现网站,最后使用内网穿透实现用户对网站的访问。

Create a handwritten digital image data set,use OpenCV image recognition technology to segment the table and extract grid pictures.The collected 100 handwritten digital forms were identified and removed,and the images after the removed forms were segmented to obtain 10,00028×28 handwritten digital images as a data set.The LeNet-5 model network is used to detect the data quality of the data set,and the test set score of the self-made data set is 98.30%.From the output results of the confusion matrix and classification report,it is found that the data set is available.Use HTML to build the website,Flask framework to realize the website,and finally use the intranet penetration to realize the user's visit to the website.