High Resolution Image Reconstruction with Enhanced Refinement Network (ERN) for Orange Fruit Disease Analysis

Due to their significant role in agriculture and global trade, the health of orange crops is of great importance. However, disorders such as orange greening are poses severe threats to orange production and farmers' livelihoods. Hence, the early and accurate detection of these diseases are vital to the production of quality fruit as well as productivity. Quick and dependable diagnosis allows the farmers to resort to control measures in time to help safeguard the yields. The present study provides an Enhanced Refinement Network (ERN) model developed for the upgradation of low resolution images is and used for the accurate detection of diseases in orange fruits. The results obtained from the proposed ERN on the orange fruit greening disease dataset and are as follows: PSNR values being 30.92, 33.674, and 35.965, respectively. SSIM values record 0.8563, 0.8948, and 0.9253, respectively and classification accuracies of 99.47, 98.64, and 97.53% when upscaling factors of 2, 4 and 6 are used.