UAV Photogrammetry-Based Accident Assessment Road Condition Analysis Using Image Classification

Abstract

Every year, many motorcycle riders die in traffic accidents around the world. In Thailand, the same is true here. The research creates an artificial intelligence model of the road situation to sentinel motorcycle drivers based on unmanned aerial vehicle and image classification. Based on unmanned aerial vehicle equipment to collect urban road condition in Thailand, an artificial intelligence model is constructed using RGB data and DSM data using image classification algorithms. The research took the supervised method by experts to label pictures. Image classification algorithms are taken to distinguish the level of road situation risk, including squeezenet1_0, AlexNet and etc. After finishing the model setting, this research compares mainly by analyzing five factors such as ACCU, SENS, SPEC, MCC, and AUC. Consequently, the DSM model is not so good for analyzing road situations as the RGB model. And the better way to analyze the DSM model is AlextNet, and in the RGB model, the squeezenet1_0 is more suitable for explore. This research's goal is to create an image classification model for illustrate in view of road situations around Thailand.