High-Precision Plant Pathology using ResNet101 for Detection of Stem Canker and Dieback in Dragon Fruit

The detection of plant diseases, especially those affecting economically important crops like dragon fruit, is crucial for ensuring agricultural sustainability. Stem canker and dieback are among the most detrimental diseases affecting dragon fruit, which can lead to significant yield losses. Current disease detection methods often rely on manual inspection or basic imaging techniques, which are time-consuming and prone to errors. Moreover, existing systems often fail to deliver early-stage disease detection, leading to delayed interventions and crop damage. In this work, a deep learning approach using ResNet101 is proposed for the automated detection of stem canker and dieback in dragon fruit. The methodology overcomes the limitations of traditional methods by leveraging advanced image processing and convolutional neural networks (CNNs) to accurately classify healthy and diseased plants. By utilizing large-scale, high-quality datasets—such as the DragonFruit Healthy Vibes and DragonFruit Disease Alert— the proposed approach enables robust model training that can distinguish between various disease stages. Key features of this work include efficient data preprocessing, such as image augmentation and normalization, to ensure model generalization across diverse environmental conditions. The ResNet101 architecture, with its residual connections, facilitates better feature extraction, making it ideal for disease detection in complex agricultural settings. Model evaluation metrics, including accuracy, precision, recall, and F1-score, are employed to assess performance and ensure reliable real-world applicability. This methodology provides an automated, scalable solution to early disease detection, empowering farmers with timely insights to protect crops and enhance productivity.