Abstract: In modern precision agriculture, early detection of fruit diseases is crucial to preventing postharvest losses and maintaining quality yields. Traditional detection methods relying on manual inspection are inefficient and prone to error. We present a real-time fruit disease identification approach that leverages the YOLO (You Only Look Once) object detection architecture as its foundational framework., enhanced with preprocessing techniques like Finite Impulse Response (FIR) filtering for image quality improvement. YOLO’s convolutional neural network (CNN) architecture enables multi-scale feature extraction, allowing effective identification of diseased regions without compromising image resolution. We present the complete workflow from data acquisition, preprocessing, and training to evaluation and deployment. Experimental results demonstrate the model’s effectiveness with high accuracy, precision, and recall, suggesting strong potential for real-time application in agricultural environments.

| DOI: 10.17148/IJIREEICE.2025.13509