LumiScan Training Framework enables easy training of neural networks for computer vision tasks, including object recognition. It also allows the user to train an existing network to adapt it to new data, or to train a network from scratch and convert it to ONNX.
Features
Retrains neural network models
Converts neural network to ONNX
Generates an inference that can be loaded into LumiScan AI Object Matching Feature
Works both on Windows and Linux
For correct and successful training, it is necessary to configure all relevant parameters in the JSON settings in advance.
Prerequisites
CUDA Toolkit 11.8 or higher
Code Meter 7.60 or higher
NVIDIA GeForce RTX 2080 SUPER
Anaconda
Linux or Windows 10 or higher
Wibu Dongle for Encryption Mat. Nr. R911427110
Documentation
Support
âšī¸ HD Vision Systems Customer Service
âī¸ +49 (0)6221 6721905
âī¸ HD Vision Systems Customer Service 1)
Related Links
The Company
HD Vision Systems supports manufacturing companies in all aspects of Machine Vision.
With its unique light field-based technology, their systems offer customers fast and easy access to automation: the light field makes it possible to reliably detect complex and shiny surfaces or fiber composites. The flexible systems consisting of sensor and software are also highly adaptable for small quantities. Each vision solution from HD Vision Systems is supplied together with a vision controller mounted on the edge. The system can then be used flexibly for various applications: In addition to generating high-precision RGB-D point clouds of detected objects and object recognition, they also provide accurate object localization and inspection.
1) Note on data protection
The email is sent directly to the partner company! This company is responsible for the processing of the data contained. Please read their data protection information before sending this email! Bosch Rexroth is not responsible for the use of the partner company's support.
