I did everything as mentioned

#1

why is my accuracy -104% ???

#2

hey @itzsakshi26gupta ,
can you please provide me a link to your code file , so that i can have a look at it.
Thank You :slightly_smiling_face:.

#3

https://drive.google.com/drive/folders/1byveuKL6MARb6U192mNV7_WQma3ZjgAZ?usp=sharing

#4

hey @itzsakshi26gupta
there are some changes that need to be made :

  1. The resize function takes shape as first height and then weight. So change that to ( h, w )
  2. In np.stack , your current shape is ( 3, something ) and you need ( something,3 ). So to do that , just add axis = 1 , as parameter to np.stack. It will do it.
  3. If now you plot your image ,you will notice that the mustache is a bit higher over nose , just because out cascade had detected that position. So , just add some more pixels in y value , to get at the correct position.

If you do these steps in the correct manner , then you will be able to achieve a score of around 95.
For 100 , you need to very small changes in mustache and glasses images , but i would suggest you to leave that , as its just a headache.

I hope this would help you.
Thank You. :slightly_smiling_face:.

1 Like