Computer vision-based year-round growth monitoring and assessment for tomato plant (Solanum lycopersicum) with pre-harvest maturity grading

• Main Author: De Luna, Robert G.
• Format: text
• Language: English
• Published: Animo Repository 2020
• Subjects: Computer vision; Agricultural innovations; Image processing; Electrical and Computer Engineering
• Online Access: https://animorepository.dlsu.edu.ph/etd_doctoral/1415; https://animorepository.dlsu.edu.ph/context/etd_doctoral/article/2468/viewcontent/De_Luna_Robert_Redacted.pdf
• Institution: De La Salle University

The agriculture system is rapidly evolving due to the increase in demands for human consumptions. Since then, botanists and farmers begun to research and explore innovative methods to improve the ideal harvest periods and production gains of crops. In this field of work, invasive and destructive techniques are still in practice, to analyze and study plants. This leads to a laborious time of work and the destruction of the crops. Machine vision and image processing techniques have become useful in the increasing need for quality inspection of fruits, particularly, tomatoes. This study deals with the design and development of a computer-vision monitoring system to assess the growth of tomato plants in a chamber by assessing its height, and leaf color and by detecting the presence of flowers and fruits. The system also provides maturity grading for the tomato fruit. Two pre-trained deep transfer learning models were used in the study for the detection of flowers and fruits, namely, the Regional-based Convolutional Neural Network (R-CNN) and the Single Shot Detector (SDD). Maturity classification of tomato fruits are implemented using the Artificial Neural Network (ANN), K-Nearest Neighbors (KNN), and the Support Vector Machine (SVM). Evaluation results show that for the detection of flowers and fruits, the overall accuracy of the R-CNN is 1.67 % for flower detection and 21.23 % for the fruit detection while SSD registered 100% and 94.77 % for flower and fruit detection respectively. In the machine learning for maturity grading, SVM produced the training-testing accuracy rate of 97.78 %-99.81 %, KNN with 93.78 %-99.32 %, and ANN with 91.33 %-99.32 %. For the selection of the final model to be used in the system, all optimized models are tested using 90 separate samples with maturity equally represented. SVM got a reliability score of 95.56 %, KNN with 90 % and ANN with 65.56 %.