2
$\begingroup$

I have a dataset which has two very similar classes (men wrestling, women wrestling). I've used InceptionV3 as a classifier to solve the problem of classifying this dataset. Unfortunately, the accuracy of this classifier doesn't hit more than 70%. Is there any suggestion about how I can overcome this problem or any other similar problems?

$\endgroup$
  • $\begingroup$ How much data do you have? $\endgroup$ – Makintosz Nov 2 '19 at 8:56
  • $\begingroup$ About ten thousand. Regardless of amount of data, I would be grateful for any suggestion $\endgroup$ – Arashsyh Nov 2 '19 at 9:54
1
$\begingroup$

If you want to classify data with similar characteristics, it would often helps if you hand craft features. For classifying women or men wrestling, you may want to try using cv2 to track human faces and feed that to your CNN as input. Example: https://realpython.com/face-recognition-with-python/

If the data is not images, you may want to do a analysis to see which feature have no clear relationship with whether it is men or women wrestling and remove them. Example: https://towardsdatascience.com/a-feature-selection-tool-for-machine-learning-in-python-b64dd23710f0

Hope that I can help you and have a nice day!

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.