My bachelor's thesis covering the topic of copper ore grains classification based on thermal images. The project includes data analysis and feature extraction from footage acquired using active thermography approach. The copper grains are characterised by tracking the number of spots with lower thermal emissivity. Classification algorithm utilizes a neural network developed with Keras interface. The document can be viewed here.
Feel free to check out the project repository, to examine the implementation.