0

I want to use CustomVision.ai to detect pack of cigarettes within a shopping cart. I have tons of images of contents of shopping carts without cigarettes, and several images of the content with cigarettes. All images are taken from the same angle, like 1 meter above the cart.

So my question is, should I rather use CustomVision.ai with Object Tracking to define the area of the cigarettes or should I use a classifier to identify just all images with cigarettes, since I could also use tons of Negative examples here. The resolution of these images is 640x480px.

Here is a sample image: enter image description here

Stef
  • 644
  • 1
  • 8
  • 25
  • Does all the cigarettes packs have the same design? – Nicolas R Feb 26 '20 at 13:06
  • From my experience with Custom Vision, I would use Object Detection. You will need a lot of samples with packs to get a correct precision and recall... – Nicolas R Feb 26 '20 at 13:17
  • No, they look different. I'd say it is like 10-20 different packs. But yeah, I am currently collecting images now with cigarettes and try out the Object Detection. Thanks – Stef Feb 26 '20 at 16:54
  • Have a good luck. In any case, if you start by object detection and want to test classification later, you can still use Custom Vision API to copy the images and transform the bounding boxes into image tags. I may create a github repo for that kind of process in the next few weeks (I already have it but needs to clean it to be released publicly) – Nicolas R Feb 26 '20 at 20:59
  • That would be awesome. Would be cool if you could give me an update once you are ready. Cheers! – Stef Feb 27 '20 at 08:42

0 Answers0