Detection of roadside vegetation using Fully Convolutional Networks

• Published in: Image and vision computing Vol. 74; pp. 1 - 9
• Main Authors: Harbaš, Iva, Prentašić, Pavle, Subašić, Marko
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.06.2018
• Subjects: Convolutional neural networks; Deep learning; Image analysis; Roadside maintenance; Vegetation detection; Deep learning; Image analysis; Vegetation detection; Convolutional neural networks; Roadside maintenance
• ISSN: 0262-8856; 1872-8138
• DOI: 10.1016/j.imavis.2018.03.008

Vegetation detection is a common procedure in remote sensing, but recently it has also been applied in robotics as an aid in navigation of autonomous vehicles. In this paper, we present a method for roadside vegetation detection intended for traffic infrastructure maintenance. While many published methods use Near Infrared images for vegetation detection, our method uses images from the visible spectrum which allows the use of a common color camera on-board a vehicle. Deep neural networks have proven to be a very promising machine learning technique and have shown excellent results in different computer vision problems. In this paper, we show that Fully Convolutional Neural Networks can be effectively used in a real-world application for detecting roadside vegetation. For training and testing purposes, we have created our own image database which contains roadside vegetation in various conditions. We present promising experimental results and a discussion of encountered problems in a real-world application as well as a comparison with several alternative approaches. •Results show the best overall per-pixel accuracy compared to other methods.•Increase in the True Negative Rate for objects similar in features to vegetation•Different types of vegetation are accurately detected regardless of illumination.•Efficient detection of vegetation near and far away from the camera 1