Advanced trainings

This advanced training course dives deep into image classification based on deep learning. Using the MVTec Deep Learning Tool, you will learn how to solve a classification problem on your own.

It takes about 50 minutes to complete the course.

Target group

This advanced training course is designed for learners with a basic understanding of deep learning who are interested in developing expertise in image classification tasks.

Objectives

After finishing this course, you will:

• know what image classification is.
• understand that data is a critical part of deep learning applications.
• have trained and evaluated a deep learning classification model.
• be ready to start your own deep learning project.

This advanced training teaches you how to use MERLIC Communicator plug-ins to communicate with external devices in your automation process.

We are focusing on the following MERLIC Communicator plug-ins:

• REST plug-in
• OPC UA server plug-in
• MQTT plug-in
• TCP socket plug-in

It will take you 100 minutes to complete the entire course.

Target group

This advanced training is aimed at

• Process engineers wanting to learn how to integrate MERLIC, with a basic understanding of system integration but limited knowledge of communication protocols.
• Automation engineers with prior knowledge of communication protocols who wish to learn how to integrate MERLIC into their processes.

Objectives

After finishing this course, you will:

• be able to evaluate which of the communication protocols supported by MERLIC suits your automation process,
• know how to integrate your MERLIC vision application in the automation process, and
• be able to control your MERLIC vision application using one of the communication protocols covered in this course.

This advanced training covers HALCON's shape-based matching along with its generic shape matching operator set. You will learn the benefits of shape matching, how to apply shape matching to your use case, and how to parameterize the most common parameters for optimal matching results.

It will take you 115 minutes to complete the course.

Target Group

This training is aimed at advanced HALCON users who want to use the shape matching technology for their applications.

Objectives

Upon completion of this course, you will:

• gain an understanding of what 'matching' is in the context of machine vision and where shape matching can benefit your application.
• get to know the workflow of generic shape matching.
• understand the most important parameters, that affect robustness and runtime.
• learn how to further process the shape matching results.
• be able to solve your first basic shape matching applications.

This advanced training introduces you how to work with 3D Surface Matching. You learn the basics of 3D Surface Matching, the whole workflow of it, and details that improve the performance.

It takes about 85 minutes to complete the course.

Target Group

This training is aimed at advanced HALCON users who has already attended course "Introduction to 3D", and who wants to do 3D object localization.

Objectives

After finishing this course, you will …

• know what Surface-Based Matching is,
• know the workflow of Surface-Based Matching in HALCON and understand what each step does,
• be able to debug the Surface-Based Matching results,
• be able to improve the Surface-Based Matching regarding speed, robustness and accuracy.

Data is crucial for any deep learning application. In fact, even the most sophisticated Deep Learning model does not help you at all without suitable data.

This advanced training will teach you what to look out for and what to avoid when curating your Deep Learning dataset.

It takes about 50-60 minutes to complete the course.

Target Group

This training is aimed at new and intermediate/advanced Deep Learning users who want to find out more about the importance of data in Deep Learning applications. Completing the “Introduction to Deep Learning” course is a prerequisite.

Objectives

Upon completion of this course, you will …

• understand why data is so important for any deep learning application.
• be able to properly and accurately define the problem you want to solve with deep learning.
• be able to assess the quality of your dataset.
• know the requirements on datasets for the different deep learning methods.
• be able to curate your own dataset and expertly evaluate the quality.

This advanced training introduces you to MVTec's Deep OCR (Optical Character Recognition). You will learn how to use Deep OCR in different scenarios.

It takes about 60 minutes to complete the course.

Target group

This training has a basic part that is aimed at new HALCON users who want to start using Deep OCR and an advanced part for users (beginners or advanced) who are interested in knowing more in depth on how to use this technology. 

Objectives

After finishing this course, you will …

• know what OCR and Deep OCR is.
• be able to use Deep OCR in HDevelop in standard and non-standard scenarios.
• understand in detail how Deep OCR detection and recognition work.
• be able to optimize the results of Deep OCR to best match your task.

This advanced training shows how to use MVTec's deep learning semantic segmentation to divide an image into regions based on predicted classes. You'll learn to label data, train, and evaluate a model for inference.

It takes about 60 minutes to complete the course.

Target group

This advanced training is aimed at HALCON and MERLIC users who want to use deep learning to segment images pixel-wise according to their class labels.

Objectives

After finishing this course, you will:

• understand what you can do with semantic segmentation,
• apply the workflow of semantic segmentation in the compatible MVTec products,
• know how to label, train and evaluate a semantic segmentation model, and
• be able to integrate the trained model in HALCON and/or MERLIC.

This advanced training shows you how to identify unknown defects using MVTec's proprietary deep-learning-based Global Context Anomaly Detection (GC-AD). You learn how to prepare your data, train, and apply a GC-AD model to your application. 

It takes about 140 minutes to complete the course.

Target group

This advanced training is aimed at HALCON and MERLIC users who want to use deep learning to find defects and detect anomalies in their application.

Objectives

After finishing this course, you will know ...

• ... what Global Context Anomaly Detection can be used for,
• ... the workflow of GC-AD in the MVTec products,
• ... how to train and evaluate a GC-AD model, and...
• ...how to integrate the trained model in HALCON and/or MERLIC

This advanced training introduces you to the development of custom tools for MERLIC to extend the existing MERLIC tool library.

In practical exercises you will learn step by step how to create a custom tool in HDevelop (with HALCON) and how to integrate the custom tool into MERLIC.

You will also learn how to debug the code of a custom tool remotely from MERLIC in HDevelop.

It takes about 130 minutes to complete the course.

Target group

This training is aimed at advanced MERLIC users who want to develop their customized tool for MERLIC to extend the standard tool set provided by MERLIC.

Objectives

After finishing this course, you will:

• know what custom tools are and where they can be used,
• understand how a custom tool is structured,
• be able to program a custom tool of varying complexity, an
• be able to debug your custom tool to find and fix bugs.

This advanced training shows you how to optimize your deep learning applications in terms of inference runtime and memory requirements using MVTec’s AI Accelerator Interface (AI²). You will learn about the recommended workflow in MVTec products, as well as important parameters and how to use them for your application.

It takes about 80 minutes to complete the course.

Target group

This advanced training is aimed at HALCON and MERLIC users who want to improve the inference speed of their deep learning application.

Objectives

After finishing this course, you will:

• know which hardware can be used via the AI² Interface in MVTec products.
• understand important aspects that must be considered for designing your system and workflow for optimal inference performance after roll out.
• be able to optimize your deep learning models for inference via the AI² Interface in HALCON and/or the Deep Learning Tool.
• be familiar with using an optimized deep learning model for inference in your application with MERLIC and/or HALCON.

In this advanced course, you will learn how to perform hand-eye calibration using HALCON. You'll explore the mathematical foundations, setup types supported by HALCON, relevant hardware, and best practices to avoid common mistakes and improve calibration results.

It takes about 135 minutes to complete the course.

Target group

This advanced training is intended for users who want to work with a vision-guided robot system and therefore need to accurately calibrate a robot to a camera system. 

Objectives

After finishing this course, you will:

• have a broad and robust understanding of the principles behind hand-eye calibration.
• know the different setups of hand-eye calibration that are supported in HALCON.
• have gained experience in hand-eye calibration with both a stationary camera and a moving camera.
• be able to troubleshoot and optimize the process of a hand-eye calibration.

This advanced course provides in-depth information on using GigE Vision cameras with HALCON. You will learn the best practices for setting up and troubleshooting your GigE Vision image acquisition setup.

It will take you approximately 120 minutes to complete the course.

Target group

This advanced training is designed for ...

• engineers looking to implement image acquisition using GigE Vision cameras with HALCON.
• engineers wanting to enhance the robustness and data throughput of their GigE Vision image acquisition.

Objectives

After finishing this course, you will...

• understand the fundamentals of the GigE Vision standard.
• grasp the basics of the part of Ethernet infrastructure that is relevant for GigE Vision.
• know how to connect and configure your GigE Vision cameras with HALCON.
• be familiar with common ways of adapting your GigE Vision image acquisition.
• be able to identify, avoid, and solve common problems in image acquisition and their causes.

This advanced training course covers the integration and use of the HALCON library in software projects written in C, C++, C#, Visual Basic .NET, and Python. You will learn how to set up such projects and access HALCON’s functionality in your chosen programming language.

Following the introductory module, the course consists of separate modules for each interface.
These modules are independent of each other, allowing you to focus on the languages that are of interest to you.

It will take you 90 minutes to complete the entire course.

Target group

This advanced training course is designed for programmers who want to use the HALCON library for machine vision tasks in their software projects.
Projects can be written in C, C++, C#, Visual Basic .NET, or Python.

Objectives

After finishing this course, you will:

• Know how to use the HALCON library within the programming language(s) of your choice: C, C++, C#, Visual Basic .NET, and Python.
• Know how to create code via direct programming and HDevelop program export.
• Know the language interfaces with their classes and data types.
• Be able to create, build, and deploy executable applications that use HALCON.