Development of Explainable Deep Learning Models for Robot Vision

The project aimed to create a series of explainable models of  Deep Learning which will:

• combine a high level of learning performance with explanatory feedback
• help human users understand, trust, and manage future autonomous systems more effectively

Using the developed explainable models, the new deep learning systems are able to explain their operation, characterize their strengths and weaknesses, and allow the user to easily understand how to operate and apply inference. In addition, the optimization process is now more easier for the user as it is simpler to detect errors and imperfections in the models.