Text recognition – various tasks for FotoNow with potential savings in the millions
In addition to automated image capture and error detection, a third area of expertise has emerged for FotoNow: The recognition of various text information, including connection and information transfer to company specific databases. An overview based on three completely different requirements situations in the automotive industry.
Information extraction from order slips or screens
Which components must be installed where? In large companies, such information is sometimes communicated digitally via screens to the employees in the assembly department, but is also provided classically on paper. On so-called order sheets, for example, it is noted whether darkened glass must be installed in a vehicle, whether it is a model for a specific market and thus automatically requires special equipment, but also things like body colour or fog lights. FotoNow can get this information easily by connecting to the company’s own vehicle information systems.
But what if the customer doesn’t want this connection and still wants to know whether a vehicle should have so-called anti-sun glasses? Then the first stage of our text recognition system comes into action. FotoNow reads this information from a screen or order sheet and then uses error detection to compare whether the correct glass has been fitted. In the event of incorrect installation, an automatic error notification would then be sent to the responsible employee. Even things like vehicle identification number, component identification and much more can be recognized in this way and, among other information, saved directly in the file name of the image.
Component tracing with database connection
A more complex form of text recognition is necessary in departments that have to deal with the issue of conformity in production. They must ensure that exactly the right component ends up in the right vehicle. With FotoNow, photos are taken of these components and the text information printed, glued or engraved on them is transferred directly into the company-specific database. This ensures that no wrong component ends up in the vehicle and therefore there can be no problems with approval for certain markets.
The challenge for our FotoNow team here was to be able to recognize the variety of differently placed information on the most diverse forms. Starting with metal surfaces, to weakly embossed numbers, to part numbers attached to a cable. A total of over 500 different components were integrated, which are photographed with FotoNow by the employees directly during the mounting process. Depending on the department, this is done using with permanently installed cameras, but also with smartphones or tablets.
Part number comparison in real time
In the After Sales area, damaged components keep turning up. These are already delivered defective by the suppliers, which often ends in a recall action. The customer is then providing lists of the affected components including their part numbers, which can be up to 32 digits long. Finding these parts quickly degenerates into a Sisyphean task, especially if the lists contain several hundred entries. The results are concentration problems for the employees due to this abundance of tasks, the time- consuming component search and the matching of numbers between paper printout and component.
FotoNow helps here by transferring the reported part numbers into our FotoNow architecture. Our text recognition then enables automatic comparison between the photo taken of the component and the information provided on the part number. In no time at all, the employee receives feedback as to whether the photographed component is a reported defective part or whether it is OK.
Increasing efficiency and saving costs
Use cases such as component tracing or the rapid location of damaged parts lead to savings in the millions for our customers. Processes are 100% guaranteed and enable carefree work, including time savings and increased efficiency in all aspects.