THREE APPLICATIONS OF DEEP LEARNING ALGORITHMS FOR OBJECT DETECTION IN SATELLITE IMAGERY

Detection of objects in images has been long used in computer vision applications (image and video analysis) in fields such as surveillance or robotics. The last decade saw a break-through in this area when deep convolutional neural networks were introduced, in addition of the GPU computing capacity. In remote sensing, satellite images are also used for feature extraction and often classic machine learning techniques are used for the classification of the pixels in the image. This paper shows how one of the networks developed for the ImageNet challenge can be applied to satellite imagery for object detection using three examples: roads, palm trees and cars.