Gidel, a technology leader in high-performance FPGA systems development, introduces a small and robust embedded computer tailored for high-throughput image acquisition and image processing. FantoVision 20 allows image processing, compression, and recording of video streams with up to 20 Gb/s in real-time, all packed in an ultra-compact form factor.
High-speed image acquisition and real-time processing
With 2 x 10GigE and Camera Link 80-bit (DECA) connections, FantoVision 20 provides high-bandwidth camera interfaces to capture and process high-resolution images at fast frame rates in real-time.
FantoVision 20’s distinct architecture combines an Nvidia Jetson Xavier NX embedded computer with an Intel Arria 10 FPGA. This unique combination features out-of-the-box high-end frame grabbing with real-time AI and image processing. Algorithms are easily programmed using C/C++ CUDA and NVIDIA’s extensive AI, image processing and compression libraries. The FPGA offers powerful pre-processing and high-end compression capabilities. Using multi FantoVision 20s and Gidel’s open-FPGA InfiniVision acquisition flow, a 1000+ cameras streams can be captured and synchronized simultaneously.
With a 134 x 90 x 60 mm3 form factor, FantoVision 20 is an ultra-compact computer that can be easily integrated in systems with limited space. Its robust hardware has been designed to operate under harsh industrial conditions. Passive and active cooling options are available.
New possibilities for industrial, outdoor, and mobile applications
With its unique processing capabilities and small form factor, FantoVision 20 opens new possibilities for developers of vision systems with high-bandwidth real-time AI and image processing requirements. Potential applications include high-speed inspection, ITS (Intelligent Transportation Systems), broadcast, medical devices and embarked applications such as agriculture, aerial mapping, etc.
With its high-speed interfaces and real-time processing, FantoVision 20 enables inspection or sorting tasks at high resolution and high frame rate. Using a 10Gigabit Ethernet switch, multiple GigE cameras can be networked for 360° inspection or 3D scanning. Machine learning and inference processing can be implemented using Nvidia’s AI libraries. Thanks to the small form factor and compression capabilities, portable recording systems can even be designed for offline measurements.
For surveillance or traffic monitoring applications, FantoVision 20’s compression, local recording and streaming capabilities combined with its compact and rugged hardware that can easily be installed on gantries or traffic lights, offers a powerful yet cost-effective solution. The supported interfaces allow for long cable lengths between cameras and the FantoVision computer.
The small form factor, light weight, and low power consumption of the FantoVision20 also makes it the computer of choice for any system embarked on a vehicle. For example, it is perfectly suited for vehicles capturing 360° views of their surroundings at high-resolution for mapping purposes or for aerial imaging. Another potential field of outdoors application is agriculture, such as for automatic weed extraction, targeted fertilization, or harvesting by vision guided robots.
With its high-bandwidth interfaces, powerful image processing capabilities, compact size, and modular options, FantoVision20 is the ultimate computer solution for high-speed and high-resolution machine vision applications requiring real-time processing, compression, or recording at optimized overall system costs.
