Automatically enhancing image quality in machine learning training dataset by using deep generative models

A system for automatically enhancing the quality of images in the training set of a neural network NN, the system comprising:

• a data storage including a plurality of images;
• an image quality assessor configured to identify, using an image quality assessment method, an image from the plurality of images in the training set matching a low-quality criterion as a low-quality image;
• an image quality enhancer configured to;
• use an image enhancement method for enhancing the low-quality image to obtain an enhanced image and replace the low-quality image with the corresponding enhanced image in the training set, and
• check the training set for presence of low-quality images;
• and
• wherein the image quality enhancer is further configured to enhance the low-quality images and replace the low-quality images with enhanced images;
• wherein after execution of the image quality assessor and the image quality enhancer is repeated a predetermined number of times, if low-quality images are present in the training set, automatically removing the low-quality images from the training set;
• and
• a sufficiency evaluator configured to identify if the training set is sufficient to train the neural network NN based on at least one sufficiency criterion after the removal of the low-quality images from the training set.