-2

I've been searching for iris recognition implementations. I just see Matlab implementations only. why python doesn't have while python has more extensive and mature libraries?

How can I create my own iris-recognition system? can you help to define the road map?

Haseeb Ayub
  • 25
  • 3
  • 10
  • Welcome to Stack Overflow. Please take the **tour** (https://stackoverflow.com/tour) and read the information guides in the **help center** (https://stackoverflow.com/help), in particular, "How to Ask A Good Question" (https://stackoverflow.com/help/how-to-ask), "What Are Good Topics" (https://stackoverflow.com/help/on-topic) and "How to create a Minimal, Reproducible Example" (https://stackoverflow.com/help/minimal-reproducible-example). – fmw42 Apr 28 '20 at 21:44
  • ok the question is extremely broad! what recognition? image, features? what is the libraries you what to use?! (opencv/tensorflow/pytorch) is there any hardware limits?! (CPU/GPU/mobile device/...) how much do you know about each will also help the details of the road map – a-sam Apr 28 '20 at 22:42
  • @a-sam **Thank you for your reply.** 1- I want to extract the features of an eye so I can distinguish b/w to personals. 2- I'm very proficient in OpenCV and tensorflow. 3- I'm currently using my laptop but in the future, I'm may be port it to the raspberry pi. – Haseeb Ayub Apr 29 '20 at 11:15

1 Answers1

1

Now that we know the problem:

The first step is to find implement-ready codes, i recommend you to take a look at these:

  1. G6-iris-recognition python package
  2. iris-recognition by thuyngch in github (python)
  3. papers with code about iris-recognition

then if the codes above did not help (which is highly unusual), you have to make these steps:

  • first search the available papers in this field, see the designs

  • if you like to implement the pipeline via neural networks

    1. train a CNN network in Tensorflow
    2. port it into OpenCV for compatibility with every device, take a look at this link for example

  • if you like to develop your algorithm via hand-customed algorithms, i highly recommend following a paper, this way is richly researched in the literature and is mature.

hopefully this helps.

a-sam
  • 481
  • 3
  • 8