In application development, there is not only the ONE AI solution. And a particular AI vision system is not always the suitable answer to a problem. The application decides how, what and with what. Simplicity, cost, performance and time belong to the most important and decisive criteria for the use of AI vision technologies. The fact that AI-based methods work completely differently than rule-based approaches is both an advantage and a disadvantage. However, this in turn enables manufacturers like IDS to develop entirely new types of tools for image processing that can be used much more intuitively. With them, human quality requirements can already be transferred to AI-based image processing systems through machine learning in order to optimize and automate processes. Often, not a single line of source code needs to be written in the process, making AI vision suitable for entirely new target groups that no longer need to have programming skills. Companies are no longer necessarily dependent on programmers and image processing experts at this stage.
Despite the intelligent new methods, AI vision remains "image processing". To realize a solution, it has to be integrated into an application, like its rule-based counterpart, and work together with other system components. Any user, with or without prior knowledge of AI, image processing or application programming, must also solve this task. And therein often lies the real challenge of an application. Anyone who selects system components purely on the basis of price, performance and richness of features often has to deal with various service providers and manufacturers in the further stages with regard to interfaces of the different systems. Getting them to work together smoothly in a complete workflow often requires patience, time and a lot of in-house expertise.
IDS therefore relies on components that are optimally matched to each other and can be used by all user groups with the support and experience of one single contact partner. This saves valuable time, expensive development resources and gets the user to his individual solution faster.
Ready-to-use embedded AI vision systems
Artificial intelligence running directly on intelligent industrial cameras - with the IDS NXT complete vision system, AI vision is easy and cost-effective to use. Individually trained neural networks for object detection, classification or anomaly detection as well as industry-standard interfaces and communication protocols turn the autonomously operating cameras into fully-fledged inference systems. The entire development process from image to finished AI application is covered by easy-to-use tools. The cloud-based AI vision studio IDS lighthouse is used to manage image data sets, train neural networks and create complete image processing sequences as vision apps for IDS NXT cameras.
The IDS NXT ecosystem will be expanded this summer with another fast and easy-to-integrate embedded vision platform. The new IDS NXT malibu reduces inference times by up to twenty times compared to the existing models - which means it can execute neural networks much faster. Machine vision results here are integrated into the live image as overlays in a hardware-accelerated manner and provided as compressed video streams (H264 or MJPG). This makes it the ideal application focus for process monitoring or for visual assistance systems, which you can see for yourself at automatica in Munich (booth B5.203).
AI vision for everyone
However, a stand-alone camera is not always the most suitable platform for a machine vision application. For various reasons, the use of a PC-based system in conjunction with classic industrial cameras may be necessary or advantageous. Performance, connectivity, flexibility, integration into existing systems, to name but a few. Ultimately, the requirements of the planned application alone determine which advantages of a platform are best suited for implementation. However, as a user, you are often faced with the situation of having to decide between different machine vision components for a particular system - with all their advantages and disadvantages. Unfortunately, since you can't cherry-pick and put together your own, you are often forced to make compromises.
IDS customers have no restriction in their choice of system with AI vision. Neural networks trained in the cloud based AI vision studio IDS lighthouse can be used with standard industrial cameras in PC-based systems in addition to IDS NXT AI cameras. Since version 2.4 of the IDS peak software, IDS AI-Vision can also be used by application developers in their own camera applications. IDS provides virtually a machine vision construction kit with all the tools through which the simplicity of AI-based image evaluation can be used with any IDS camera, i.e. not only those of the IDS NXT product line. This allows AI-based inference to be added to any application already running that uses IDS cameras.
IDS has expanded its AI vision portfolio in a short period of time, providing users with the appropriate component for the required platform for each application. From the low-cost "It's so easy!" full featured embedded vision system for simple applications without PC up to high-end image analysis with PC and industrial cameras, everything is covered. Users benefit from seamless support from the camera to the image-based training of a CNN and integration into their own application.