Deep learning features for 3D object recognition

Halcon version 25.05 combines deep learning algorithms with classic methods to improve industrial image processing.

© MVTec

A key new feature is Deep 3D Matching, which was developed for bin-picking and pick-and-place applications. This technology enables fast and robust 3D object recognition and position determination based on 2D images. Users can train their own 3D matching models without external support. In addition, existing functions such as the code reader for QR codes and deep OCR have been optimized. Halcon 25.05 also offers a new image acquisition interface for modern camera technologies and an updated version of H-Develop EVO with new features such as AI assistants and a gray value histogram. These improvements aim to increase the flexibility, performance and usability of the software.