An Approach to Facilitate Business System by Multiple Barcode Detection using Faster RCNN

Barcoding system is a cheap and reliable way of tagging the products. The barcode detection process is needed for an inventory system to detect the barcodes of the products and for the billing system of the products. Nowadays, laser scanners are used to detect single barcode in super shops, but they are costly. If multiple barcodes could be detected from an image, it may help everyone to save some more time than scanning them separately. In this paper, a model that has been proposed to develop which will work better for detecting and decoding multiple barcodes simultaneously. In this work, the Faster RCNN model is used for the detection of multiple barcodes. The detection process is also done using the Pyzbar library separately. But Faster RCNN gives us better output to detect barcodes from an image than this library does. With the help of TensorFlow API, we worked on our dataset for the detection process of barcodes using transfer learning method. The decoding process is done on the Arte-Lab dataset with the help of the Zbar library. Though the detection rate using Faster RCNN is a little bit slower, but it gives better accuracy. The detection and decoding accuracy throughout a model can facilitate a business system for faster transaction.