AI vision in just a few steps

Image processing can be so simple

Break the rules - Think Inferent

Image processing with IDS NXT is based on system components that optimally work together and a step-by-step workflow that leads to your own application in a way that is easy to understand.

The big advantage over classical embedded vision development with rule-based image processing is the simplification. Tasks that require special knowledge in embedded programming or neural network training are handled automatically by IDS NXT. Users can thus concentrate on the essentials. The complete system provides support in a few phases from the idea to the complete running application.

“It‘s so easy!”- Workflow
Step by step to your application

IDS NXT offers a simple approach to reach the goal in all project phases. Get to know the IDS NXT components.

1 Collect image data

IDS NXT allows easy collection and import of image data for different image processing solutions from various sources via a simple file repository.

Application images can originate from an existing database, be conveniently prepared via the IDS NXT cockpit and IDS NXT cameras, or simply be captured with a smartphone.

2 Organise datasets

IDS lighthouse allows the import of images from simple folder structures with simultaneous assignment of categories for classifications or anomaly analyses.

In addition, an ROI editor makes it possible to selectively mark image areas for object detections.

The data management allows all application data to be optimally prepared both for training and for associated tests of your image processing.

3 AI Training

Run one or more parallel trainings with different image and test data, or alternative specifications at the push of a button and get informed about the training status by email.

Training results can be analysed and validated directly in the web protocol with your test images using tools like the confusion matrix and attention maps. This allows neural networks to be specifically improved and retrained before they are deployed.

4 Create your Vision App

With the block-based Editor and the application wizard, individual image processing sequences can be created using machine learning methods, such as detection or classification of objects, even without programming knowledge. The applications can also be tested and debugged with your image data sets in the cloud, even before they are loaded and executed on an intelligent IDS NXT industrial camera.

5 Run AI Vision App on-the-edge

Vision apps created with IDS lighthouse can be easily uploaded and activated on an IDS NXT camera via the IDS NXT cockpit. Vision apps, e.g. for classification or object detection, are also available on the cameras to execute your trained CNNs.

Maintain and extend applications

Vision Apps created with the block-based Editor can be easily further edited via the camera's web frontend.

With new insights and further image data, you can continuously improve your neural networks with the AI Vision Studio IDS lighthouse and thus increase the throughput of your application.


You need help with an application or have specific questions about getting started with IDS NXT?
We have prepared various information to help you quickly.