The critical shortage of on-site radiologists in rural or emergency settings often leads to delayed diagnoses of orthopedic injuries, which can result in improper bone healing or long-term complications for patients. Current diagnostic workflows frequently rely on cloud-based processing, which is impractical in low-connectivity environments or during transit in ambulances. This project addresses these gaps by developing a real-time computer vision system deployed on an edge device, such as a specialized tablet or embedded processor, to detect bone fractures directly from X-ray images. Utilizing a lightweight, optimized deep learning model like YOLO or MobileNet, the system provides instant binary classification and visual localization of suspected fractures without requiring an internet connection. This ensures high data privacy by keeping patient information local while acting as a critical decision-support tool for first responders and general practitioners. Ultimately, the project aims to reduce clinical fatigue-related errors and significantly accelerate the triage process in high-pressure medical environments.