Visual obstacle detection for UAV

Abstract

Recently, a great deal of computer vision's most innovative and state-of-the-art object detection algorithms have evolved around deep learning. With the rise of Deep Learning (DL) from Machine Learning (ML), it has emerged among the greatest technological advancements and inventions in the advancing age of our technological inventions. In the context of Deep Learning (DL), Convolutional Neural Networks (CNN) are regarded as one of the most critical components. Recognizing images and detecting objects is something that CNN has achieved significant success in. Nonetheless, CNN can be very large in size, and it carries an extremely high load of logical computations. As a result, a new type of CNN, called You Only Look Once (YOLO), was developed to detect and classify objects. Additionally, it provides a smaller overall architecture and faster computing capabilities.