
Real-Time Detection of Terrain Surface Types and K-Factor for Path Profiling
Software as a Service
NoCategory
Thesis DesignClients
Confidential
Techstack
PythonYOLOv7OpenCVPygameDJI Tello
Purpose
An automated path-profiling study that classifies terrain surface types in real time from a drone camera feed using YOLOv7, computes the corresponding k-factor for each surface, and evaluates the overall accuracy of the system.
Description
A DJI Tello drone streams live video to a Python application that runs YOLOv7 inference through PyTorch, classifying terrain as landforms, bodies of water, or man-made structures. A Pygame-based interface presents detections in real time while the system looks up k-factor values from a labeled JSON mapping and logs captures, records, and detection runs for accuracy evaluation. OpenCV handles frame processing between the drone feed and the model.




