IDS NXT rio & rome

Inference cameras for industrial purposes

The two IDS NXT camera families are characterized by their highly flexible application spectrum. They are fully-fledged standard industrial cameras whose range of functions can be extended and modified by users as required thanks to vision apps. Due to their powerful hardware and software, they are ideally suited as embedded solutions - in this case, image processing is partially or completely performed on the cameras ("on the edge"). This reduces bandwidth utilization and saves space and costs for an additional PC computing unit. With their special AI core, the industrial camera families are also able to execute neural networks very efficiently.

Thanks to the app-based principle, these cameras are as versatile as smartphones. Vision apps allow users to quickly and flexibly determine which image processing task should be performed - for example, reading codes, characters or license plates, recognizing faces, anomalies or scratches in surfaces, or finding, measuring, counting or identifying objects. With the help of a neural network it is also possible, for example, to classify parts into "good" and "faulty" parts. Recurring vision tasks can also be set up and changed quickly.

Tailor-made solutions can also be implemented very easily using the app development kit. Users can create their own individual vision apps in just a few steps and install and execute them on IDS NXT cameras. Depending on the app, the camera delivers results or signals directly to machine controls or supports PC-based applications with pre-processed data. "Smart GenICam "* offers even more advantages in the future: even camera functions programmed by the user itself can be queried and executed by third party applications in compliance with GenICam. Nevertheless, the devices remain fully-fledged standard-compliant industrial cameras that can transmit image data at full Gigabit Ethernet speed.

Inference cameras in the housing of standard industrial cameras

Neural networks that have been trained by users for individual tasks can be executed directly on an IDS NXT camera using the integrated AI core "deep ocean core". This turns the device into a powerful inference camera that makes artificial intelligence usable in an industrial environment. Image analyses then take place locally, thus avoiding bandwidth bottlenecks in transmission.

The FPGA-based AI acceleration enables inference times of a few milliseconds. IDS NXT cameras can keep pace with modern desktop CPUs when it comes to the accuracy and speed of AI tasks - while at the same time consuming considerably less space and energy. The re-programmability of the FPGA provides additional benefits such as low recurring costs and time-to-market. With the dissemination and networking of such "cyber physical components", direct process data is made available according to the principle of the "Internet of things" (IoT). Beginning with the digitization of plants up to the provision of status information through predictive maintenance, they enable the automation and efficiency increase of industrial manufacturing processes.

*available in the next development stage