About Dataset Upper-Extremity Bone Fracture Segmentation & Detection (X-ray) A comprehensive dataset of upper-extremity X-ray images was prepared for automated bone fracture localization and classification in computer vision research. The primary goal of this dataset is to support the development and benchmarking of algorithms for fracture detection and pixel-level fracture segmentation on radiographs. The dataset includes X-ray images grouped into multiple fracture-related categories covering common anatomical regions of the upper limb. The provided classes (as defined in data.yaml) are: Elbow Positive Fingers Positive Shoulder Fracture Wrist Positive Humerus Forearm Fracture Humerus Fracture Each image is annotated with pixel-level segmentation masks (polygon-based), enabling precise localization of fracture regions. These mask annotations can also be used to derive bounding boxes, making the dataset suitable for both instance segmentation and object detection pipelines (e.g., Ultralytics/YOLO segmentation training and evaluation). Dataset Highlights Total images: 3,715 Train: 3,316 | Validation: 257 | Test: 142 Total annotated instances: 4,294 Annotation format: YOLO segmentation labels (normalized polygon coordinates) Application scenarios: fracture region localization, multi-class fracture recognition, detection-to-segmentation cascades, and clinical decision-support prototypes. By providing diverse fracture categories across upper-extremity regions, this dataset aims to facilitate robust model development for real-world radiographic fracture analysis, accelerating research in computer-aided diagnosis and improving clinical workflow efficiency.