Fruit-vegetable detection using mask region convolutional neural network for harvesting decision

Mohd Noor, Nur Natasha (2022) Fruit-vegetable detection using mask region convolutional neural network for harvesting decision. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Preview

Text (Full Text) Fruit-vegetable detection using mask region convolutional neural network for harvesting decision.pdf - Submitted Version Download (2MB) | Preview

Abstract

In recent years, deep learning has shown excellent capabilities for learning image features and has been widely used in object detection applications. One of the applications in deep learning is agriculture, where traditionally, farmers used their experience to determine the types and quality conditions of fruits or vegetables in their farms. A new technology known as smart agriculture or precision agriculture was introduced to solve this problem. A type of deep learning known as Mask Region Convolutional Neural Network (Mask R-CNN) has become the state-of-the-art fruit detection method. Hence, this method was explored in this project. With the advent of deep learning techniques, fruit/vegetable detection accuracy can increase drastically by choosing the best backbone architecture for the system. Three types of backbone architectures have been analyzed, namely VGG-16, mobileNetv2 and ResNet101. ResNet has achieved the best result with 96% accuracy. A Graphical User Interface (GUI) was also developed as a system's guide to the user. A precise detection result can aid the farmer in crop harvesting decision at the right time, according to the fruit or vegetables of choice, especially for business.

Actions (login required)

View Item