Unlike municipal waste, medical waste poses greater potential risks if not treated properly, especially in the initial classification stage. Misclassification can lead to increased infection hazards and environmental harm. To address this issue, this study develops an automated detection and classification model for medical waste with the aim of supporting accurate identification during the early sorting phase. The method employed is You Only Look Once version 11 (YOLOv11). We used a dataset of more than two thousand manually labeled images collected across diverse waste types. The results indicate that the model delivers accurate recognition and strong overall performance, with both Precision and Recall exceeding 95%, while the mean Average Precision (mAP) metric demonstrates stable detection capability across multiple Intersection over Union (IoU) thresholds. By enabling faster and more reliable waste classification, this research provides a practical and environmentally friendly solution that reduces infection risks, supports healthcare workers, and contributes to safer and more sustainable medical waste management. In addition, the pa-per compares the three models, YOLOv9, YOLOv10, and YOLOv11, thereby confirming the strong applicability of YOLOv11 in medical waste classification.