A Review on Deep Learning on UAV Monitoring Systems for Agricultural Applications

In this chapter we present literature review on UAV monitoring systems that utilized deep learning algorithms to ensure improvement on plant and animal production yields. This work is important because of the growing world population and thus increased demand for food production, that threaten food security and national economy. Hence the need to ensure sustainable food production that is made more complicated with the advent of global warming, occupational preference for food consumption and not food production, diminishing land and water resources. We choose to consider studies that utilize the UAV platform to collect images compared to satellite because UAVs are easily available, cheaper to maintain, and the collected images can be updated at any time. Previous studies with research foci on plant and animal monitoring are evaluated in terms of their advantages and disadvantages. We looked into different deep learning models and compared their model performances in using various types of drones and different environmental conditions during data gathering. Topics on plant monitoring include pest infiltration, plant growth, fruit conditions, weed invasion, etc. While topics on animal monitoring include animal identification and animal population count. The monitoring systems used in previous studies utilize computer vision that include processes such as image classification, object detection, and segmentation. It aids in increasing efficiency, high accuracy, automatic, and intelligent system for a particular task. The recent advancements in deep learning models and off-the-shelf drones open more opportunities with lesser costs and faster operations in most agricultural monitoring applications.