Building application for recognizing svg/vectors from camera

For the development of a new product, we require an application to recognize images from live feed camera (usb/eth). These images must be compared to a SVG file and identify the product it is scanning. when identified, a database must be accessed to get info on this product. The camera will be placed on top of a conveyor belt, so we can insert product (flat). The products will have a maximum dimension of 150x300cm. The camera view will be - depending on the height to the converyopr belt - slightly distorted,so we must have the option to calibrate the camera to a raster (e.g. 10x10cm). Some products will have the same outer dimension, but are mirrored or with more holes. Some products are small and there will be multiple products on the conveyor belt at the same time. This should be tackeld by the application. The whole application should be web based, with the possibility to put it in a android application in the near future. The output consists of the (1) identified product and (2) the location of the product on the conveyor belt (x,y) at the moment of identification. The development will be done in 3 stages: stage 1: access local camera and calibrate it to raster; displaying raster stage 2: compare image to svg and identify + positioning (e.g. using opencv) stage 3: front end, and database connection