AT logo - The Machine Vision Specialists  
  Frame Grabbers  
  Vision Systems  
  Contact AT  
  Site Map  
Request product information


(08) 9242 5411

(02) 9905 5551

(03) 9384 1775

Defence Recognised Supplier Scheme Logo


Deep Learning Tool: a custom-made classification bonus for HALCON and MERLIC software users

Go to Machine Vision Software
Go to MVTec

Deep learning: complementary technology solution
for machine vision classification applications

What is "Deep Learning"
Deep learning is a concept that uses multiple layers to transform input data into a slightly more abstract representation. For example, the first layer may identify edges; the second layer may identify the arrangement of edges; a third layer may identify a facial feature and a fourth layer may understand that the image contains a face.
Deep learning can discern which elements to place, optimally, in which layer, on its own.

Deep learning is used in a wide range of industries, as well as for machine vision purposes, for example, speech recognition on smart phones or in social media to identify a user's friends in uploaded photographs. However, deep learning solutions are specific to a particular need and are not easily transferable to a different application and face-recognition deep learning software can't be used to identify faults in an electronic wafer - each application requires its specific solution.

Augmenting pre-trained classification networks with "Deep Learning"
Traditionally, when working with machine learning, the user was required to select, manually, the features that most effectively separated the classes. This complicated task does not exist when using deep learning, as, during the training, the network automatically chooses which features to use, the user merely providing a quantity of labeled images.

Deep Learning in your application
Developed to work perfectly with HALCON and so ensuring seamless integration, MVTec's Deep Learning tool provides labeling functionality for HALCON's deep-learning-based, axis-aligned object detection.

Labeling training data is a critical factor in any deep learning solution and with the deep learning tool, labeling is a simple operation. By drawing axis-aligned rectangles around each relevant object, and by providing information about their corresponding classes, users can easily train their own classifier

HALCON and MERLIC both include deep learning OCR and HALCON offers a seamlessly integrated, comprehensive set of other deep learning functions

Read more about HALCON and MERLIC

Classification Semantic Segmentation Object Detection (axis-aligned) Object Detection (oriented)
Deep learning-based image classification easy assignment of images to trained classes.

Requiring little effort results in very short set-up times and applying the classifier to new data is especially fast.
Because trained defect classes can be discerned with pixel, accuracy, inspection tasks which were previously either unsolvable or solved by expending considerable programming effort.
Object detection localizes trained object classes and identifies them with a surrounding rectangle.

Touching or partially overlapping objects are also separated to enable counting. The rectangles can also be aligned according to the orientation of the object, resulting in a more precise detection, as rectangles then match, more closely, the shape of the object .

The deep learning tool enables the labeling of training data needed for drawing axis-aligned rectangles and the resultant HALCON dictionary can then be integrated, seamlessly, into HDevelop in order to train the network. Helper scripts are provided for this functionality, and for other deep learning tasks.
Workflow when using MVTec's Deep Learning Tool

Need a price or more application information? Please email Adept Turnkey or call our offices

Adept Turnkey Pty Ltd is"The Machine Vision and Imaging Specialists" and distributor of MVTec in Australia and New Zealand.
To find out more about any Teledyne DALSA product, please call Adept Turnkey at Perth (08) 9242 5411 / Sydney (02) 9905 5551 / Melbourne (03) 9384 1775 or contact us online.




If you like this page, please recommend and share it.

Google+ Facebook Twitter More