Looking at QR Codes

This is relatively straight forward, make a virtualenv and install `opencv-contrib-python`. (For running in Lambda, we'd use the `-headless` version.)

Note that you do not have to otherwise install opencv. The pip install will get everything you need for these to work.

import cv2

To get started, we'll first make a QR code and see what it looks like. I do make it larger, but that is just so I can see it easily.

qr_encoder = cv2.QRCodeEncoder_create()
code = qr_encoder.encode("This is just a silly example.")
code = cv2.resize(code, (100, 100))
cv2.imwrite("silly-qr-code.png", code)

And reading that code back in?

qr_detector = cv2.QRCodeDetector()
qr_detector.detectAndDecode(code)[0]

This is just a silly example.

I decided to hunt around my house and see if I had any codes around. For reasons of weak will power, I had some chips. So, snapped a photo.

And then quickly got disappointed that I couldn't decode it.

food_label = cv2.imread("food-label-qrcode.jpg")
qr_detector.detectAndDecode(food_label)

('', None, None)

I tried running the image through some filters, but never managed to get this to detect. Left me feeling pretty frustrated.

Luckily, I saw that there is another decoder in the contrib directory. I originally didn't want to try it, as I was not sure I could get it running in an AWS Lambda. In particular, I don't want to use docker, so was limited in file size.

That said, it does the job.

wechat_qr_detector = cv2.wechat_qrcode_WeChatQRCode()
wechat_qr_detector.detectAndDecode(food_label)

(('http://pepsico.info/490k32',), (array([[ 506.0724 , 1682.5837 ],
       [1346.1771 , 1644.7062 ],
       [1340.8032 , 2498.914  ],
       [ 489.23245, 2523.8735 ]], dtype=float32),))