Lung diseases refer to diseases or disorders that affect the lungs and prevent them from functioning properly. The utilization of Artificial Intelligence in predicting lung diseases has proven advantageous in the medical field. This paper classifies the detection of lung diseases using Deep Residual U-NET (ResUNET), a Deep Learning Semantic Segmentation technique. The model was developed using the U-NET architecture in conjunction with a residual neural network. Binary Cross-Entropy classification facilitated efficient training, optimization, and comparison. The model’s performance was enhanced using metrics such as Jaccard Index, Intersection over Union (IoU), Dice Coefficient, and accuracy. A dataset of chest X-rays and their masks was used as input, with the predicted masks being the evaluated output. Our model attained a 98.10% accuracy rate, proving its efficacy in lung illness identification. By using skip connections inside residual blocks, our ResUNET model efficiently propagates information and extracts features. This method is a potent tool for medical image analysis since it uses deep learning algorithms for binary segmentation, yielding competitive results with slightly greater accuracy than existing cutting edges techniques.