Green Chili Pepper Localization System Based on Mask R-CNN and IoT

Chili is one of the most widely grown crops and is highly consumed in Myanmar. However, the monitoring and harvesting of chili in Myanmar are still carried out manually by human labor, consuming a significant amount of time and energy. Additionally, plant growth can only be monitored on-site. The integration of harvesting robots and IoT techniques has become crucial for the advancement of modern agriculture. IoT based system can view the situation of chili field with the help of pi camera. The color of ripe chili, stem and branches are the same color as background leaves. So, it is difficult to localize stems and branches on each ripe chili when harvesting time. This is a highly challenging task for harvesting and monitoring operation in the field for vision-based to carry out. Solving the issue, this paper developed the Mask Region-Based Convolutional Neural Network (Mask R-CNN) by combination of IoT gateways. The proposed system comprises two stages: identifying ripe or unripe fruits and localizing branches. The main objective of the study is to develop a method that can locate the location of the cutting area of chili branches. In order to verify the model's performance, it was further evaluated on a testing dataset, where it yielded 92.7% precision and 94.8% recall.