Digital object detection from high resolution satellite data using generic advance deep learning tools and methods
Dr. Tabassum H Khan, Pranay Nakhale, Rohan Lanjewar, Sakshi Nandanwar, Isha Gujar ,
The Automaton Farm Boundary Detection Platform is designed to provide an accurate, scalable, and cost-effective solution for identifying and analyzing agricultural farm boundaries in high-resolution satellite imagery. The platform leverages deep learning models, particularly YOLOv8, CNN, and DenseNet, to detect various objects essential for crop monitoring and protection. The process begins with the input of high-resolution farm images into the YOLO v8 modal. The images are first resized to match the model’s input size and then undergo pixel normalization to standardize the data for processing. The images are passed through deep learning algorithms, which extract relevant features from the data, helping to identify key objects within the farm, such as crop types, boundaries, and other elements crucial for analysis. Keywords—Object Detection, Feature Extraction, Image Classification, High-Resolution Satellite Imagery, YOLO, Edge Detection.