Batch number, production date and price information are printed at a specific location on Product lables in a factory environment and they come out in a conveyor belt. It is required to detect any lables which don't get printed with any data (a blank space where the printing should be there) or with erroneous or unreadable printing.
A hardware device, perhaps a Raspberry Pi with a suitable camera attached to it will be mounted above the conveyor belt. From the video stream, the erroneous labels should be identified and signal (a beep) as they are moving.
The two condition of erroneous labels are as follows:
1. No printing at all in the box
2. Text printed in three lines with Price, Month & Year of packaging and batch number (as shown in attached image) should match exactly with the data text entered at the start of the program. If there is a mismatch or unreadable/smudged text, program should indicate (beep) as error label.
A picture of sample blank label, a printed label and a video of lables coming out in the conveyor belt is provided. Any other information or more videos can be provided. It is required by the developer to generate sufficient test data, train the model and provide a python code which can be executed on Windows or Linux environment. Loading the code to Raspberry Pi environment is not the responsibility of the developer and will be handled by ourselves.
11 freelancers are bidding on average ₹43005 for this job
Hi We need more than 500 images of each with and without label parcels. I will build and share model then which can classify labeled images and images with no label. Thanks