The independent electronics design house has already produced over a hundred thousand computer vision devices, with the main solutions being driver tracking cameras while driving, automatically tracking situations on the road, etc.
The development team is confident that the development of a computer vision device begins with the selection of solutions for optics. The selection of the perfect combination of sensitive elements (matrices) and lenses serves as the foundation for subsequent design. An error in the number of megapixels or an incorrect viewing angle of the optics can play a sick joke on further development, stealing a valuable resource for rework, revision, and hacks.
“At DANNIE group, before beginning the design of any camera, we conduct a large-scale analysis of the business tasks and functions of the future device: the installation location, the required range and quality of the shots taken, the end user's operating conditions,” says Viktor V., head of Computer Vision Systems. “This enables us to determine the optics of the device being developed at an early stage while taking economic and technological factors into account".
If the client's objective is to identify large objects, such as cars and obstacles on the road, a matrix with a lower resolution can be used. At the same time, a high-resolution matrix is required to track more detailed objects, such as products on a retail shelf. It is critical to consider all of the details here. An unfavourable outcome of selecting the incorrect optical node can be a multiplication of labour costs for software development. Of course, this will have a significant impact on the project's overall cost.
“Each of these tasks is a fascinating engineering challenge", Victor is certain.
DANNIE is currently implementing several major computer vision projects while constantly improving its internal expertise.
One of the company's advantages is the established network of contacts with manufacturing factories that produce layouts of optical modules as close as possible to serial samples in the shortest possible time.