Robotic based Pneumonia Diagnosis and Prediction using Deep Learning with Dropout Regularization Technique

The automatic detection of pneumonia in medical images is a major and serious process because hard to identify the illness in medical images. We have observed that many people are dying every year due to pneumonia. The objective of the defined work recommends a better elucidation for this delinquent, with a new deep learning classical prototype. This novelty consists of dropout technique is embedded as a part of convolution neural network. The planned model is trained and tested with 6217 images. The chest X-ray images were selected from many patients, suffering from illness. The outcomes of our proposed model are gives better results compare to existing convolution neural network models. The model is analyzed with different statistical metrics, it gives 97.22 accuracy, 97.45% precision, 97.35% recall. The new model uses different dropout rates, the overfitting is reduces with 40 % dropout rate and gives better accuracy, recall, precision and Area Under the Curve. These results are acceptable and competent with current deep learning models.