Hard to decipher?
Deep learning will do it!
The challenge: Barely readable characters
The solution: Deep Learning and VC MIPI®
MITECH Group: Finding and reading engraved characters
In this handheld edge-device, the VC MIPI® IMX296 camera takes images of the objects. Deep learning algorithms on the NVIDIA Jetson nano processor board find and read engraved characters (alphabet and 0-9) and output the results on a display. The device will be used to recognize numbers and letters engraved in metal parts. Especially in outdoor applications and installed over several years in humid and dirty environments, these characters are often barely readable for the human eye and thus time-consuming to find and identify. Deep Learning helps to recognize the characters quickly and provides reading results including confidence-values. This reduces the risk of faulty assignment.
The image shows a demo-setup from an early prototyping stage. VC is a NVIDIA Preferred Partner and provides MIPI cameras, drivers and accessories for the quick and easy setup and development of embedded vision solutions with NVIDIA and MIPI. The NVIDIA processor boards of the Jetson TX2™, Jetson Nano™, Jetson Xavier NX™ and Jetson Xavier AGX™ platforms are characterized by their performance, industrial-grade quality and long-term availability. Our VC MIPI® camera modules also share these features and are in combination with the NVIDIA boards a perfect fit for any embedded project. With low power-consumption and ultra-compact design both processor boards and VC Cameras are ideally suitable for any mobile application and edge devices.
Barely readable characters
VC Vision Solution
VC MIPI® IMX296 for NVIDIA Jetson