Machine learning algorithms and Computer Vision technologies for detecting 52 playing cards deck

Question

https://en.wikipedia.org/wiki/Standard_52-card_deck

https://en.wikipedia.org/wiki/Playing_card

https://www.google.com/search?q=playing+cards&client=ms-android-lava&prmd=sinv&sxsrf=ALeKk02cmtbG25vrqk59BybOJWHF7PG3Aw:1600237953234&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjcuIj2hu3rAhXtzjgGHZegCj0Q_AUoAnoECAwQAg

https://www.google.com/search?q=playing+cards&client=ms-android-lava&prmd=sinv&sxsrf=ALeKk02cmtbG25vrqk59BybOJWHF7PG3Aw:1600237953234&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjcuIj2hu3rAhXtzjgGHZegCj0Q_AUoAnoECAwQAg

https://www.instagram.com/p/CFMBjc6J7PN/?utm_source=ig_web_button_share_sheet

Are there Machine learning algorithms & Computer vision technologies which can view the 52 playing card deck?

Input : 52 playing card deck images as .jpg,.gif,.tiff

Output : Detecting the playing card names.

For example : Spade Ace, Three diamonds, Jack of Hearts, King of Clubs.

All the playing cards names should be detected by inputting playing cards images.

If no, what are the limitations? Do the 52 playing card images have to be modified?

One can also add one more playing card as "Joker" to the existing 52 playing cards added as a image as input.

Answer 1

Yes, you can recognize these cards. For easy implementation, you can check here and here. Also, you can build your custom neural network model with tensorflow, keras, pytorch etc. Recognizing visually "52 card deck" is already solved problem. Because, cards have good features & landmarks. You either can use neural networks or "old school" computer vision algorithms.