Our Model detects whether the X-ray image is of patient diagnosed by Pneumonia having a Consolidated lung or a healthy person with Normal lung.
We have analyzed our data on 2 different models one with Efficient Net and other by applying attention over Efficient Net.
Our dataset contains two classes namely Consolidation and Normal which in total contains 2043 images . Consolidation class has 1016 images. Normal class has total 1027 images.
Earlier Model without using Attention had accuracy of approx 49% but after using Attention accuracy significantly jumped to approx 65%.