Convolutional Neural Network based on Transfer Learning for Medical Forms Classification 34
Adil Fahad Alharthi · International Journal of Advanced Trends in Computer Science and Engineering · 2019
Guaranteeing a high accuracy across image documents' understanding technologies is very difficult in commercial or even in experimental field.Our aim was to work in known context on documents of interest from health organisms and to hospitals' users in KSA, to offer solutions that help with the automatic or semi-automatic processing of health forms.We treated; particularly, the problem of medical forms understanding based on Convolutional Neural Networks for deep features extraction.We used transfer learning with two pre-trained architectures as AlexNet and GoogLeNet.We evaluated these features for the classification problem in different setups, using several public datasets; together with, using our proposed database with a KNN classifier.Our results showed that deep features are of high performance by reaching 94.2% of recognition for our contextual dataset.