Can an Image Classifier Recognize Different Hair Textures?

TensorFlow General Discussion
,
1

I’m working on a project that involves recognizing different hair textures () from images. I am considering using TensorFlow for this purpose.

Details
I would like to know if it’s feasible to build an image classifier that can accurately classify various hair textures using TensorFlow. Specifically, I’m interested in understanding the following aspects:

  • can the model distinguish between type 3a or 3b for example

  • What are the challenges I might face in terms of image preprocessing and feature extraction?

Thank you for your assistance!

Hair type has 4 major categories: straight hair (type 1), wavy hair (type 2), curly hair (type 3), and coily hair (type 4). These 4 types of hair are further categorized into A, B, and C subtypes creating 12 total categories:

  • Type 1A Straight Hair
  • Type 1B Straight Hair
  • Type 1C Straight Hair
  • Type 2A Wavy Hair
  • Type 2B Wavy Hair
  • Type 2C Wavy Hair
  • Type 3A Curly Hair
  • Type 3B Curly Hair
  • Type 3C Curly Hair
  • Type 4A Coily Hair
  • Type 4B Coily Hair
  • Type 4C Coily Hair
3

Hi @GwnElviin, you can build an image classification model using Tensorflow.

Distinguish between finer hair texture (like 3a vs 3b) might be challenging, but it can be possible with a dataset having images with clear variations between these types.

while performing image preprocessing you can face the problems like maintaining class balance. As the model performance depends upon the feature extraction it might be challenging for choosing the best feature extraction technique for your use case. Thank You.

1 Like