Various factors such as age, stress, injuries, and congenital issues may affect the shape of the lower limb, leading to abnormal alignment of the limb. The manual delineation and measurement of the Hip-Knee-Ankle (HKA) region on X-ray radiographs using traditional techniques may be both time-consuming and inaccurate. To automate this procedure, a variety of deep learning techniques have been developed, including angle calculation, coordinate regression, and area of interest segmentation. The Masked R-CNN approach automates lower region segmentation, identification of lower extremity regions of interest (ROI), landmark coordinate determination, and extra Hip Knee Angle Angles (HKAAs). This method has potential for aiding in the prediction of landmarks and areas of interest on the lower extremities, as well as automating the segmentation of the HKA area. The recommended approach uses new Masked R-CNN deep learning algorithms for ROI segmentation to identify and correct lower-limb misalignment caused by Orthopedic Analysis (OA). This model accurately diagnoses lower limb misalignment, setting the groundwork for efficient treatment approaches on the knee. The model meets the desired performance standards.