Semantic Segmentation

Classical thresholding and edge detection become unstable on glossy surfaces, under changing light and against busy backgrounds. A deep-learning segmentation model learns to recognise the product from labelled examples and extracts the area the product covers in every image as a mask.

This is how we find radiators on Elba's panel radiator line: even on the glossy, striped surface of the roller conveyor the model separates the radiator from the background, reaching over 99% overlap (IoU) for the radiator class on test images. The outline and position found are used by the robot to pick the product.