Deep convolutional neural network for detecting plant disease in apple leaves

Abstract

Agricultural diseases stand as a significant threat to global food security, yet their accurate diagnosis remains challenging due to inadequate infrastructure in various regions worldwide. Notably, signs manifesting on plant leaves often provide crucial indicators of illness presence. Moreover, with technological advancements, the precision and effectiveness of diagnosing both animal and plant ailments have significantly improved. The identification stage sets the foundation for a series of concerted efforts to combat and curtail disease spread. Within this paper, the YOLOv3 model, hinging on advanced deep transfer learning object detection technology, is harnessed to develop a robust system for recognizing healthy and diseased apple leaves. The study's findings reveal an impressive detection approach with an mAP value of 96.38%. Given its superior efficiency compared to preceding methods for diagnosing apple plant diseases, the proposed model emerges as a suitable solution for both apple orchards and apple plant tree producers.