<p>This research aims to develop a CNN-based computational framework for diagnosing bone fractures from X-rays, improving osteoporosis knowledge. Current osteoporosis management focuses on elderly fracture prevention, but many are unaware of their condition until a fracture occurs. This framework seeks to enable earlier diagnosis and management for all individuals, preventing fractures and improving quality of life, especially for at-risk and aging populations. These interesting findings prove modern deep learning systems as potential diagnoses for bone fractures even in cases of ones suffering from osteoporosis. This paper examines the use of DEXA radiography augmented by deep learning algorithms to estimate fracture risk beyond mineral density determination. The existence of computer-aided designs in numerous areas cuts across and helps to address the issues that are presented with the current methods of diagnosis. In the field of bone radiography, X-ray has been opted for in this study since it is more effective than CT scans. In this research, an advanced CNN model was applied that included a mix of architectural modifications to classical models and data augmentation techniques to outperform other standard CNN-based models. The Canny edge detection algorithm and the contour detection algorithm, both based on OpenCV, were incorporated and supplemented using Adam optimization. The focus of this work has been the enhancement of CNN model to detect fractures and localize it at the same time. The best model solely attained an accuracy level of 98.23%, outperforming all other tested models, including VGG16, VGG19, Random Forest Classifier, ViT’s, and its various versions.</p>