Global Context Anomaly Detection and Deep OCR Training

The new version Halcon 22.05 by MVTec Software offers numerous new features: Until now, the Deep Learning technology Anomaly Detection was only able to detect localized, structural anomalies. The new Global Context Anomaly Detection, on the other hand, is able to “understand” the logical content of an image. Only good images are required for training. Labeling of the training data is not required. In addition, an improved print quality check for ECC-200 codes is available, as well as a new operator that performs adaptive histogram equalization to locally improve the contrast in an image, and Deep OCR training, which allows users to perform individual training on their own application data set.

Image: MVTec Software GmbH
MVTec Software GmbH