beh.cam

Learning from your employees and co-workers by training an artificial intelligence. Simply capture pictures from your application and label them accordingly.

  • Capture and label pictures
  • Training of the AI in minutes
  • Optimized productions and processes

How to use our product

(Un-)supervised Learning

State-of-the-art machine learning algorithms are used to classify images, objects or pixels or detect anomalies from your trained, ideal world. Starting with less than 20 training samples for fair and easy problems.

Classification

Which object is shown in the image? To which class does it belong? What is the quality level?

Get a corresponding class to the given image, based on your training.

Examples: Sorting, action trigger, quality assessment

Detection

Which objects are in the frame? Where are they located? Extract a list of classified bounding boxes from the image and grab or sort the objects.

Examples: Object sorting, quality assessment, object count, object position

Advantages through technology

Computing Architecture

No need for expensive server farms or different components from multiple suppliers. Our integrated and scalable processing platform, highly optimized for machine learning tasks, runs with a few watts inside beh.cam.

Algorithms

State-of-the-art algorithms, also used in autonomous driving, are trained for your application. This enables high robustness, even in challenging and changing light situations, and occlusions.

Fast and easy extension

The machine learning models can be updated and extended continously, without starting from scratch. Simply add additional labeled images, in order to further improve your model accuracy or handle corner cases. Also additional classes can be added to your model.

Fast integration & intuitive usage

Our hierarchical web-based user interface allows fast and easy capturing and labeling of your dataset. Training and optimization is accelerated on the graphics processor of the web-client. Integration in your industrial environment is possible with common safe field-bus protocols.