Vision Apps – one camera, countless possibilities
One camera for many different tasks
With Vision Apps, image processing tasks on IDS NXT cameras can be executed and changed as easily as apps on a smartphone. One camera can easily perform multiple tasks, like measuring objects in sorting machines, reading codes during incoming goods inspection or capturing and pixelating faces. There are almost no limits for the users’ creativity. Even the combination of AI-based tasks, such as object detection, in combination with classic image processing is feasible. To enable customers to adapt the cameras their individual needs, IDS has released the development environment IDS NXT Vision App Creator. To ensure the shortest possible path to the first personal Vision App, the company also provides development libraries and sample apps.
The Vision App concept ensures that users are able to determine for themselves which image processing tasks their cameras solve. By using apps, the versatile IDS NXT devices are quickly configured and put into operation. Developers can use the recently released IDS NXT Vision App Creator SDK to design individual vision apps and thus adapt IDS NXT cameras even better to their requirements. This results in highly flexible devices that can be used for numerous vision tasks.
There are even more advantages of the IDS NXT platform. It offers a convenient way to implement artificial intelligent-vision tasks, for example, even if the user has no previous knowledge in the fields of deep learning or camera programming. The all-in-one AI-solution IDS NXT ocean includes all necessary tools and workflows.
Therefore, it supports the user right from the first steps with the new technology. Camera hardware, software, infrastructure and support come from a single company. Users only need sample images and knowledge on how to evaluate them (e.g. “good” / “bad”) to create a neural network. This makes the start into AI-based image processing particularly easy.
With the help of the IDS NXT lighthouse cloud software, even non-experts without prior knowledge of artificial intelligence or camera programming can train an AI classifier with their own image data. Since it is a web application, all functions and the necessary infrastructure for creating the neural network are immediately available.
This means that users do not have to set up their own development environment first, but can start training their own neural network right away. This involves three basic steps: To upload sample images, to label the images and then to start the fully automatic training. The generated network can then be executed directly on the IDS NXT industrial cameras, turning them into powerful inference cameras.
An inference camera can apply the “knowledge” acquired through deep learning to new data. This makes it possible to automatically solve tasks that would either not be possible with rule-based image processing, or would require great effort. Since IDS NXT industrial cameras have a special AI core, neural networks are hardware-accelerated and run directly on the devices – enabling inference times of just a few milliseconds. With features such as C-mount, robust housing, GigE network connection with RJ45 or M12 connectors, RS232 interface and REST web interface, they are also fully-fledged industrial cameras. The IDS NXT rio and rome models are now available as serial cameras with different sensors and protection classes. Thanks to OPC UA, IDS NXT cameras can not only be easily integrated into industry 4.0 systems, but can also directly trigger subsequent processes as AI-capable edge devices.
IDS also offers an IDS NXT ocean design-in kit which is particularly useful for anyone who wants to test the potential of AI for individual vision tasks. It provides all the components a user needs to create, train and run a neural network in his productive environment. In addition to an IDS NXT industrial camera with 1.6 MP Sony sensor, lens and cable, the package includes six months of access to the AI training software. The use of deep learning-based image processing for individual applications can thus be realized in a short time.