04.04.2014  pavel

In-house training course in Wageningen

Tomato classification from hyperspectral image' In the beginning of April, we organized an in-house training course in Wageningen UR (Plant Sciences Group). It was a great to work with this highly experienced team specializing in agricultural automation. Our course covered the entire pattern recognition system design life-cycle from problem definition, feature selection, classifier training to application-specific performance optimization and real-time deployment.

The advantage of an in-house training is a strong focus on the client application domain, in WUR case on image-based defect detection and sorting in agro and food sector. Therefore, we leveraged a number of WUR data sets - typically spectral or hyperspectral image sets of different plant/vegetable products. This gave us an ample opportunity to discuss specific approaches such as data labeling and visualization techniques, local image and spectral feature extraction or detection/classifier cascading.

On the deployment side, we've practiced with several different targets including C/C++, .Net and LabView. We were exporting the trained classifiers from Matlab and directly executing them on precomputed features or on raw input data.

Course testimonial from Erik Pekkeriet, Senior Project manager at Wageningen UR:
"Our team on Vision & Robotics choose to use perClass in our machine vision solutions on a structural base. Therefore we followed the in-house training at Wageningen UR with the whole team of (senior) software researchers. The in-house training was from a very high level, perfectly fitting the market demands. Now we should be able to develop faster with a better score on classification issues in agriculture and food applications."

In-house perClass training in industrial machine learning in Wageningen.'
In-house perClass training course in Wageningen.'
In-house training course in industrial patter recognition.'






Remember my personal information

Notify me of follow-up comments?

Submit the word you see below: