Strawberry Diseases Detection Using Adaptive Deep Residual Network

Abstract

The development of strawberries is often impacted by inorganic and genetic terms, leading to significant risks to both quality and productivity. However, the existing approaches for disease recognition are characterised by a high rate of misinterpretation. Due to the high requirement for high strawberry productivity, relying on conventional recognition techniques primarily based on personal expertise and visual inspection is insufficient to address these challenges. Hence, it has become essential to develop more efficient approaches for accurately detecting strawberry diseases, along with providing detailed disease descriptions and suitable control measures. This work presents a clustering-based Deep Learning (DL) model for strawberry disease recognition. Initially, the input images are normalised, and the affected regions are segmented by the Fuzzy C Means (FCM) clustering. Finally, the categorisation of different diseases is classified using the DL model Adaptive Deep Residual Network (ADRN). The ADRN is the integration of the Deep Residual Network (DRN) and the Reptile Search Optimizer (RSO). The analysis is evaluated on the Strawberry Disease Detection Dataset and attained better accuracy and precision of 0.991 and 0.995, respectively.