Dynamic Bird Detection Using Image Processing and Neural Network

Collisions of aircraft and birds cause serious flight accidents, and various studies are underway to find a solution to the problem. In recent image recognition studies, state-of-the-art deep learning technologies have been actively applied. This paper proposes image preprocessing and bird detection methods in all dynamic environments using Convolutional Neural Network (CNN) technology. Image preprocessing separates moving creatures from the dynamic background and removes the background. When image preprocessing is complete, the image of the moving object remaining in the frame is used as input data for the learning model to determine whether the bird is in the frame. We used the Inception-v3 neural network model to improve the accuracy of small object classifications.