Abstract Scope |
In-situ monitoring of shrinkage during sintering is a critical challenge for manufacturers of ceramic products. Lucideon Limited, a long-established materials development and commercialisation organisation, is developing a robust, non-invasive computer-vision based system, that harnesses Deep Learning and Image Processing. This system facilitates real-time monitoring of the specimen’s boundaries during the sintering; minimising the risk of specimen’s shape distortion within the kiln, a challenge often caused by poor visibility of the process. The approach has implemented and tested advanced deep learning models, such as U-Net and YOLO, alongside state-of-the-art unsupervised edge detection algorithms, facilitating boundary identification without dataset labelling. Also includes a robust image pre-processing module, all seamlessly integrated with a user-interface.
The project not only stands out for its practical impact but also promises significant cost savings for industry by reducing material wastage and optimizing the sintering process, marking a promising advancement in both image processing and material science. |