SmartBelt-BeltMisalignment

This dataset was collected as part of an industrial research project aimed at improving the safety and separation efficiency of conveyor belt systems in barrier eddy current separators. It includes a comprehensive set of images and videos captured under various real-world industrial conditions—including high brightness, low-light scenarios, and strong machine-induced vibrations—while the conveyor belt was in motion. 1- Conveyor Misalignment Detection: Contains annotated images and videos used for detecting and correcting belt misalignment through machine vision. Data were recorded under dynamic industrial conditions to ensure robustness and reliability. 2- Thermal Fire Detection: Includes thermal camera images of the conveyor belt area above the magnetic drum. These images, analyzed for abnormal heat patterns, were used to develop a fire prevention system that detects the presence of ferromagnetic contaminants causing overheating. 3- Edge Detection: Contains both raw and labeled datasets of fine materials (e.g., aluminum, copper, plastic) moving on the conveyor belt. The data are used for detecting sharp-edged versus smooth particles via segmentation and classification models. The labeled dataset is annotated with segmentation masks, while the raw dataset includes original images from a line scan camera.