MVTec brings deep learning to NVIDIA Pascal architecture
December 14, 2017
The deep learning inference in the new version 17.12 of the HALCON machine vision software was tested on NVIDIA Jetson TX2 boards based on 64-bit Arm processors.
Munich. MVTec Software GmbH has made deep learning functions available on embedded boards with NVIDIA Pascal architecture. The deep learning inference in the new version 17.12 of the HALCON machine vision software was tested on NVIDIA Jetson TX2 boards based on 64-bit Arm processors. The deep learning inference, i.e., applying the trained convolutional neural network (CNN), almost reached the speed of a conventional laptop GPU (approx. 5 milliseconds). This is possible thanks to the availability of two pre-trained networks that MVTec ships with HALCON 17.12. One of them (the so called "compact" network) is optimized for speed and therefore ideally suited for use on embedded boards. A software version for this architecture is available on request.
In addition to deep learning, the full functionality of the standard machine vision library HALCON is available on these embedded devices. Applications can be developed on a standard PC. The trained network, as well as the application, can then be transferred to the embedded device. Users can also utilize more powerful GPUs to train their CNN, and then execute the inference on the embedded system.