PILLNOW: An Integrated CNN-YOLO and KNN Approach for Real-Time Pill Detection and Classification
Year: 2026
Type: conference-paper
Abstract: This paper presents a smart computational framework for real-time pill identification using CNN YOLO and KNN algorithms. The system leverages image recognition and anomaly detection to enhance elderly medication adherence and caregiver support, ensuring safety and accuracy. Experimental results demonstrate that the CNN YOLO model achieved high precision and recall across multiple pill classes, with several medicines reaching near perfect detection performance. The integration of KNN further improved classification reliability, particularly among visually similar pills. Overall, the framework delivered strong detection accuracy, stable performance under varied lighting conditions, and low latency, confirming its suitability for real-time medication management.
Institution: Naga College Foundation, Inc.
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7166554591