Automatic Bird Identification for OffshoreWind Farms

Abstract

There is a need for automatic bird identification system at offshore wind farms in Finland. The developed system should be able to operate from onshore, which is cost-effective in terms of installations and maintenance. Indubitably, a radar is the obvious choice to detect flying birds, but external information is required for actual identification. A conceivable method is to exploit visual camera images. In the proposed system, the radar detects birds and provides the coordinates to camera steering system. The camera steering system tracks the flying birds, thus enabling capturing a series of images. Classification is based on the images, and it is implemented by a small convolutional neural network trained with a deep learning algorithm. We also propose a data augmentation method in which images are rotated and converted in accordance with the desired color temperatures. The final identification is based on a fusion of data provided by the radar and image data. We present the results of the number of correctly identified species based on manually taken images.