Paper Title
SEGMENTATION OF PNEUMONIA USING UNET AND DEEPLABV3

Abstract
Abstract - Pneumonia, a serious and potentially life-threatening condition, predominantly impacts the elderly demographic. In acknowledging the transformative potential of medical imaging, particularly leveraging advanced techniques like deep learning, this system presents a methodology for segmenting pneumonia from chest X-ray images. The core of the approach involves deploying a deep learning model designed specifically for segmentation tasks. The training process harnesses the established and highly relevant Kermany dataset, renowned in the field for its applicability in chest X-ray images depicting pneumonia. The accuracy and robustness achieved in the segmentation of chest X-ray images are crucial for the subsequent phases of diagnosis. To achieve precision and robustness in the segmentation process, advanced deep learning methods, specifically DeepLabV3 and UNet, have been employed. These state-of-the-art techniques bring a high level of sophistication to the task, leveraging the capabilities of deep neural networks. DeepLabV3 is known for its effectiveness in semantic segmentation tasks, providing detailed and accurate delineation of objects within images. On the other hand, UNet, with its architecture featuring contracting and expansive paths, excels in tasks requiring precise localization. According to evaluation results, this system achieves more accurate results with DeepLabv3 than with UNet architecture. Keywords - Deep Learning, DeepLabV3, UNet, Pneumonia Segmentation