Analysis of aerial images for identification of houses using big data, UAV photography and neural network

Computer vision has undergone significant transformation owing to deep learning in the last two decades. Deep convolutional networks have been successfully applied for various applications to learn different tasks related to vision, such as image classification, image segmentation, and object detection. Deep learning models can generate fine-tuned results by transferring knowledge to large generic datasets. This study aims to conduct an in-depth analysis of a big data tracking algorithm for aerial images of unmanned aerial vehicles (UAVs) to detect houses using neural networks to address the low accuracy and efficiency of manual detection in remote areas by mitigating the associated security risks. In the context of big data, a UAV-based preprocessing method is discussed for images using guided filtering. In order to reduce the impact of radiation distortion on the color and brightness of UAV-based aerial images of houses, a histogram matching method was applied. The guided filtering method is used to solve the problem of imaging details of houses that are not apparent after smoothing and denoising the aerial images. A house detection algorithm based on a deep neural network is then applied to the UAV images to detect the images of houses, and the time consumption of the deep learning operation is examined within the context of big data. Combining deep separation convolution and calculation optimization with YOLOv2 improves the house's image detection in real-time while preserving an accurate performance of UAV-based aerial images to detect houses by combining the YOLOv2 detection framework. The results of the experiments indicate that the proposed method can improve the efficiency and accuracy of house detection using aerial images and has certain practical applications.