Have you realized today’s life with cameras everywhere imaging nearly everything? One reason behind its possibility is chip scaling, which accommodates key components from semiconductor makers which facilitate equipment manufacturers to fabricate quarter-size cameras. These are now becoming ubiquitous in agricultural, automotive and industrial markets.
Inspecting into the camera manufacturers and smart form of image processing is what “AI to the edge” is to vendors while the machine form perceives it as a vision camera with deep learning inference capabilities. This justifies the different perceptions of manufactures, researches, and vendors.
When looking into the marketing pitch of approach into these centers and the proposition at which they are placed, there are high chances for deep neural network acceleration on a camera. This allows inferences of performance in the network, eliminating intuitions of the raw video stream for further processing.
FLIR Systems, Wilsonville, Oregon, show a combination of the Movidius stick and the software development kit which will enable acceleration into prototyping subsequently. The initial production version of the camera configures a 1.6-megapixel CMOS image processor from Sony Corp. (NYSE: SNE) for the front end and the Movidius Myriad 2 VPU for deep neural network inferences.