Detection and Localisation of Farm Mangoes using YOLOv5 Deep Learning Technique

Abstract

Fruit detection and localization in its real environment is a very challenging task. There are multiple factors such as occlusion, shadows and varying light conditions that make this task very tough. YOLOv5 object detection and localization model was used to perform this task due to its high inference speed. The model was trained on Google Colab using a 150 image dataset of mango fruits in the farm. The developed model detects and localises the farm mangoes with a mean average precision (mAP) of 0.434 and F <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</inf> score of 0.57 at a confidence of 0.876 during training and validation. The model was able to detect and localize the field mangoes with 94.3% accuracy from the test images. The model proved its adequacy for the real-time detection of the farm mangoes using the recorded video with an average accuracy of 80.76%. From the results, it can be stated that, the developed deep learning model is a useful tool for the yield monitoring and harvesting using robotics.