Redefining industrial cameras
What would smartphones be without apps? You could use them to do little more than make phone calls and send messages. IDS NXT adapts the app principle to industrial image processing: the standard functions of these new industrial cameras can be extended as required using vision apps. Artificial intelligence opens up additional application possibilities. The IDS NXT ocean end-to-end AI solution makes it particularly easy to get started. From camera hardware including self-developed AI core to intuitive training software for creating neural networks and support, everything comes from a single supplier. IDS NXT models are transformed into powerful inference cameras by installing neural networks. This makes it possible to automatically solve tasks that rule-based image processing could not solve at all or only with great effort. IDS NXT cameras thus offer added value that standard GigE cameras cannot offer - and represent a new stage in the evolution of digital industrial cameras.
The All-in-One Inference Camera Solution
Artificial intelligence for industrial cameras
Artificial intelligence (AI) opens up new fields of application for camera technology and image processing. IDS NXT ocean shows what an easy-to-use solution for industry looks like. With the IDS NXT ocean design-in kit, customers receive a complete solution that includes not only the required hardware components but also online access for intuitive training of neural networks for their own AI applications. Thanks to the cloud software IDS NXT lighthouse, even non-specialists without any prior knowledge of artificial intelligence can train an AI classifier with their own image data that can be executed immediately on IDS NXT cameras. This makes it particularly easy to get started with deep learning-based image processing.
Hardware, software, infrastructure, knowledge and support are all provided by IDS: both the training software for neural networks and the IDS NXT industrial cameras with their powerful AI core ("deep ocean core"; (FPGA technology that performs hardware-accelerated execution of artificial neural networks) were developed in-house. Users only need the knowledge about their images and their evaluation to create a neural network. This makes it possible, for example, to solve image processing tasks with strongly varying objects - such as classifying different types of fruit or identifying defective parts (e.g. apples with pressure marks or colour deviations). Describing all occurring variances with classical image processing would be extremely time-consuming and thus costly. With artificial intelligence, on the other hand, such challenges can be solved in the twinkling of an eye.
App your camera
Vision app-based image processing with the IDS NXT platform
The functions of classic vision sensors - such as barcode readers - are usually strictly defined and hardly expandable. IDS NXT shows that there is a different way: the freely programmable platform is not tied to any predefined tasks. Instead, users can develop and install vision apps themselves and then change and set up various image processing tasks in a short time. App development does not require any additional specialist knowledge, because it hardly differs from the development of a "classic" industrial camera application. The possibility of installing different types of vision apps creates universal application possibilities in numerous application areas, for example as an analytical device in medical technology, for monitoring tasks using face recognition or vehicle and person counting. The cameras also show their strengths in optical quality assurance - for example in AI-based classification tasks by distinguishing "OK" / "NOK".
Moving applications from the host PC to the camera
All IDS NXT models allow image processing "on the edge", so that even pre-processed data or final results can be transmitted to PCs or machine controls. Depending on the required performance, users can decide for themselves whether their application should be run classically on a PC workstation, independently on the IDS NXT camera, or collaboratively on both devices. IDS NXT therefore moves image processing away from the PC and closer to the image source, making image acquisition and processing "on the edge" industry- and vision-compliant available in industrial cameras. The models are thus no longer mere image suppliers, but can also work completely independently, trigger subsequent processes directly or support a connected PC through individual pre-processing. Thanks to the embedded solution, the load on the host PC, energy consumption and system costs can be reduced.
Outlook: Smart GenICam
Use self-programmed camera features GenICam-compliant *
When users want to execute self-created functions on intelligent cameras, they face two challenges. Because there are often only proprietary interfaces available, processing data in third-party applications becomes more difficult, plus few models offer a high-speed data interface like GigE Vision. However, industrial applications often require both.
The solution are IDS NXT industrial cameras with Smart GenICam: the user uploads the desired new camera function as a vision app on an IDS NXT model. Its firmware then integrates the application code into the camera's range of functions. Control and function parameters as well as results of the self-programmed application are then made available in standardised form via the device-specific XML description file of each GenICam-compliant application. Third party applications then use these features as if they were part of the standard functionality of the industrial camera.
* available in the next development stage