Drones are constantly evolving and are set to provide cost-effective solutions to crucial real-world problems. The remote-controlled drones which were previously used for military purposes are now being equipped with sophisticated sensorial devices for data acquisition, and algorithms are being developed for autonomous flights. In parallel to the upgrades on drones, other fields such as real-time analytics and deep learning algorithms are also experiencing drastic and positive enhancements. The combination of these technologies opens the doors to potential novel architectures and provides solutions for unsolved problems to date. This chapter focuses on the weed detection branch of precision agriculture, and four different architectures are proposed. These architectures are geared towards real-time imaging data acquisition from agricultural fields, processing of the images, analytics, extraction, detection and relaying of insights. The main contribution of this work is the proposal of an in-flight hybrid architecture based on that of Apache Spark platform to support the components of the deep learning algorithm. Battery lifetime is a well-known constraint of UAVs. However, the speed of operation of Spark’s framework enables the smooth implementation of the proposed in-flight hybrid architecture with a reduced impact on battery life and flight time. The proposed architectures have the potential to facilitate large-scale farming and decrease the use of herbicides.
Drone-Based Weed Detection Architectures Using Deep Learning Algorithms and Real-Time Analytics
Algorithms for Intelligent Systems
14.03.2022
19 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Drone Object Detection Using Deep Learning Algorithms
IEEE | 2021
|Towards Real-Time Drone Detection Using Deep Neural Networks
Springer Verlag | 2021
|DRONE INSPECTION ANALYTICS FOR ASSET DEFECT DETECTION
Europäisches Patentamt | 2020
|