This study comprises a computer vision-based detection system for ten specific medications, selected due to their high sales volume and significance in the treatment of common conditions within Colombia. To address this, a YOLOv8 model was central to the developed solution. Data collection and preparation were meticulous, involving images captured in both controlled and commercial environments. This rigorous approach was designed to guarantee the system’s robustness against real-world variations in lighting and packaging, which are common challenges in practical settings. The classification system relies on the YOLOv8 architecture, chosen for its notable precision and speed in object detection and classification qualities essential for real-time application functionality. Our development workflow was further streamlined by integrating tools such as PyTorch, Ultralytics, Roboflow, and Comet for efficient model development and management. Model training involved careful hyperparameter tuning to strike an optimal balance between accuracy and computational efficiency. During testing, the model demonstrated robust capabilities in identifying these medications, achieving an impressive 94% precision and 97.2% sensitivity. Performance evaluations on both GPU and CPU platforms yielded inference speeds of 8 and 71 milliseconds per image, respectively. While the inference speed naturally decreased on CPU, the model maintained its high precision and sensitivity. This crucial finding underscores the model’s adaptability and functionality even on devices with limited computational resources, significantly broadening its practical applicability.