1. Home
  2. Products
  3. AI-Workstation

Are you ready for democratizing AI?

Have you been wondering what the excitement surrounding Computer Vision is all about? Were you thinking about taking the first steps in the field of Artificial Intelligence and testing the waters of AI? Start the journey by developing projects for future-orientated complex AI applications with the unique AI-Workstation powered by Robotron & Kontron!

Creating AI models has never been easier

You don't need any coding skills, complicated software installations, cloud applications or hardware connections. All you need is the new AI-Workstation Base System to create an AI model using the Robotron RCV Software Bundle. The hardware is already optimized for the use of the Realtime Computer Vision (RCV) solution and ready for immediate use.

 

Realtime Computer Vision for everyone:

Suitable for beginners and advanced users

Independent realization of projects and POCs

Ready for use immediately

Secure on-prem solution for sensitive data

Complete package consisting of hardware with pre-installed software bundle

Intuitive no-code environment with integrated online help and detailed tutorials

Model creation does not require programming skills

Selection of pre-trained and customizable AI models

How the AI-Workstation functions

Our AI-Workstation-solution comes pre-installed and offers a no-code GUI that caters to both beginners and advanced users. The "Cockpit" GUI facilitates the import, standardization and pre-processing of images and serves as a central administrative tool for your AI projects.

 

Tutorials

RCV - Cockpit Overview

The tutorial introduces Robotron's Realtime Computer Vision (RCV) Cockpit GUI - the administrative center for the toolkit central. It explains the main features and functionalities.

RCV - Create Topic and Import Images

The tutorial shows how to import images and annotate them by defining object classes with the help of the RCV Labeltool.

RCV - Labeltool

The tutorial comprehensively explains the main functions like classifying objects, drawing bounding boxes, or performing image segmentation, and provides details and tips of navigation, adjustments and settings.

RCV - Image Transformations

The tutorial demonstrates how image transformations like scaling, cropping or filling can be realised.

AI Modell Training

The tutorial shows the implementation of an AI model training in the RCV Toolkit providing important and useful tips for performing the training and exporting the trained model.

RCV - Create Topic Additional Features

The tutorial explains how to further enhance the annotations of images with the additional features 3D bounding boxes, Tracking ID, and depth information provided by the RCV Toolkit.

RCV - Create Virtual Topics

The tutorial provides useful tips for structuring and book-keeping large datasets by using the toolkit's additional features of creating virtual topics or filter options.

RCV - Cleanup and Delete

The tutorial explains the differences between the cleanup and delete functions and illustrates how both are performed.

RCV - Copy Topic

The tutorial shows how to copy a topic and which parameters and options are available and should be considered.

RCV - Export Topic

The video explains the export functions including the most important settings for it.

RCV - Folder structure and start-up scripts

The video illustrates the folder structure of the RCV Toolkit, including details about the subfolders and its contents, as well as an overview of the logs folder Information about the required Pytorch and Tensorflow scripts is also discussed.

RCV - Trained Model and Suggest Function (Pytorch)

The video depicts the subfolders in the Pytorch folder and the steps involved to use a trained model for inference using the suggest functionality on new images.

RCV - Trained Model and Suggest Function (Tensorflow)

The video depicts the subfolders in the Tensorflow folder and the steps involved to use a trained model for inference using the suggest functionality on new images.

RCV - Known Issues on the AI-Workstation

The tutorial shows shows some of the known issues and how to resolve them.

RCV - Anonymization

Learn how to use the anonymization services to hide sensitive information.

You don't have enough images to create an AI model?

The "Cockpit" offers a collection of prevailing data augmentation scripts to enlarge the size of your dataset. Synthetic image generator tools embedded in the "Cockpit" enable you to quickly create a new dataset of the component superposed on generic backdrops or using generative models. Annotation of images in all ranges of granularity such as classifications, bounding boxes and segments can be implemented using the simple and intuitive tools available in the "Label Tool" GUI. The GUI offers drop-down menus and as well as hot-keys, thereby catering to the needs of both beginners and professionals.

Do you have a large dataset and are you feeling fatigued from all the labeling?

Take advantage of the open interfaces integrated into the "Label Tool" to connect pre-trained models and accelerate the labeling process by using the assisted-labeling feature.

The solution comes with a selection of pre-trained and customizable AI models. By selecting a data set and a pre-trained model in the "Training" GUI, you can initiate a transfer-learning process with a One-Click-Solution. 

Not sure about what hyper-parameters to use? Just let the "Training" GUI supply a preset tailored for your system. The progress of the model training can be monitored utilizing the live logs or the graphical visualization tool, “Tensor Board”. Once the training is finished, the open interfaces of the toolkit can again be used to deploy the model and perform inferences directly in the "Label Tool" or connect them to other API's.

With our AI-Workstation, creating Computer Vision based AI models is as simple as turning on your PC and opening a browser. Try it for yourself: our experts will be happy to provide you with further information!

 

Robotron & Kontron:
Creating the future of AI together

Kontron is a global leader in IoT/Embedded Computing Technology (ECT). As part of the S&T technology group, Kontron offers individual solutions in the areas of Internet of Things (IoT) and Industry 4.0 through a combined portfolio of hardware, software and services. With its standard and customized products based on highly reliable state-of-the-art technologies, Kontron provides secure and innovative applications for a wide variety of industries. As a result, customers benefit from accelerated time-to-market, lower total cost of ownership, extended product lifecycles and the best fully integrated applications.

For more information, please visit: www.kontron.com

We look forward to hearing from you!

We will be happy to answer your questions and provide you with individual advice.

Eric Barth
Head of Sales Technology and Services

Formular Kontaktperson

* mandatory field
Information about data protection and how to handle your personal data can be found here.