ARIA: Augmented Reality and Artificial Intelligence enabled mobile application for Yield and grade prediction of tomato crops

• Published in: Procedia computer science Vol. 235; pp. 2693 - 2702
• Main Authors: B V, Balaji Prabhu, R, Shashank, B, Shreyas, Jois Narsipura, Omkar Subbaram
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 2024
• Subjects: augmented reality; deep learning; tomato detection; tomato grading; tomato yield prediction; tomato detection; deep learning; augmented reality; tomato yield prediction; tomato grading
• ISSN: 1877-0509; 1877-0509
• DOI: 10.1016/j.procs.2024.04.254

Agriculture is the most crucial sector of the Indian economy. Lately, there has been a surge in the usage of technologies like deep learning and computer vision to make the process of agriculture modern and consequently, lessening the mistakes related to conventional processes. This work delivers a user-friendly, accessible, and novel approach for the detection, counting, and grading of tomatoes found on a farm. An Augmented Reality (AR) based mobile application is developed to obtain the images efficiently from a tomato farm in the pre-harvest stage subjected to open situations. The proposed approach uses Faster RCNN, a convolutional neural network model for detection of tomatoes from the input image on a large scale. The proposed model is trained and tested using 2083 images. The results are then analyzed for the overall performance of detection, segmentation, and classification of tomatoes. The results have examined the efficiency of the proposed mobile application and demonstrate the robustness it exhibits for the detection, grading and yield prediction of Tomatoes.