The interpretation of chest X-rays is crucial in the identification of cardiothoracic and pulmonary disorders. In order to avoid catastrophic harm to individuals, this time-consuming and challenging approach must be error free, quick, and trustworthy at all times. Radiologists may benefit from using a computer-aided detection system (CAD) to help them quickly and accurately interpret chest radiographs. Development and testing of deep CNN architectures for identifying lung opacity in chest radiography were among the objectives of our research. We were interested in finding out how well architects could discriminate between normal and abnormal chest X-rays in their work. Under the same circumstances, a CNN-based model (Xception) obtained an AUC of 91% and an accuracy of 83.95%, respectively. The AUC, sensitivity, and accuracy of the Xception lung opacity classification model are higher than those of the classic chest X-ray classification model (excluding the abnormal class). By analysing deep CNN classification skills, this work may be able to increase the performance of an autonomous lung opacity detection system.