Pokemon goa challenge

#1

i am training my svm for pikachu and i am using images in gray scale of 32,32 shape but my loss is not decreasing
here is my code:https://ide.codingblocks.com/s/176757
also how can this algo works for RGB images becaue RGB images will have 3 dimensions?

#2

You missed the normalization of image part, you need to divide image_data with 255.0.
I normalized it and gor the required graph, sharing the link of my ipynb file.

Also if you want to work with rgb channel, i have attached the code for the same as well, just this time you have to reshape it in 32323.

Here is the link of ipynb file.
https://drive.google.com/file/d/1qdCvG1DkAbOjhqkzEMJnvWY7bKw3jhRt/view?usp=sharing

Hope this cleared your doubt :blush: