| CPC G06T 7/0004 (2013.01) [G06V 10/42 (2022.01); G06V 10/774 (2022.01); G06V 10/82 (2022.01); G06T 2207/30108 (2013.01)] | 20 Claims |
|
1. A method for artificial intelligence-based visual quality inspection of parts manufactured on a shop floor, comprising:
acquiring a set of real images of nominal parts manufactured on the shop floor,
executing a self-supervised pre-trainer module to pre-train a loss computation neural network in a self-supervised learning process using a first dataset created from the acquired set of real images, wherein the loss computation neural network is pre-trained on pretexts defined by real-world conditions pertaining to the shop floor, the first dataset being labeled by automatically extracting pretext-related information from image metadata, and
executing a main anomaly trainer module to train a main anomaly detection neural network to reconstruct a nominal part image from an input manufactured part image in an unsupervised learning process using a second dataset created from the acquired set of real images,
wherein the unsupervised learning process comprises using the main anomaly detection neural network for processing input images from the second dataset to output respective reconstructed images and measuring therefrom a reconstruction loss to be minimized, and
wherein the reconstruction loss includes a perceptual loss that is measured by feeding each input image and the respective reconstructed image to the pre-trained loss computation neural network and computing a measure of the difference between feature representations of the input image and the respective reconstructed image at one or more layers of the pre-trained loss computation neural network.
|