An advanced AI-driven approach to burned surface quantification in fire performance testing of wood materials

Abstract

This collection contains data obtained during flammability tests of diffrent wood-based materials used to train ai models for matching burn area.The data was recorded using a test stand equipped with a burner mounted on a guide controlled by a stepper motor and a USB camera. The entire setup was controlled by an ESP32 microcontroller and dedicated software.ZIP files contains masks:mask - original CV generated masks, used to generate ai model (train_cv.pt),manual - manual corrected masks, used to generate ai model (train_manual),ai-cv - masks generated by ai model based on CV generated masks,ai-manual - masks generated by ai model based on manually corrected masks.cpp/py files (also in train_software.zip) - software versions used in data generating, testing and analysing process.pt files - generated models.png files - samples of masks.