Then don't store them in a variable! Store them in a sprite, and
Also, how do you know it would? What if we just used 100 images instead of 1000?
Can I join? I can provide storage. (I rarely store stuff to my google drive.)
All the data in a project goes to your PC's RAM.
Ye, you can join.
You see this?
If we used 1000 GB for the project it would first, kill our personal cloud storage and second be super laggy.
Machine learning has to have a LOT of images, in this context 1000 is small.
100 GB of storage.
Again:
Also how would we upload 1k images.
2k 1k of each.
I mean we have to do each photo individually.
or use a script to do it,,,,,,,,,,,,,,,,,,,,
That's ridiculous. Just use 100 or maybe 200.
yeah, I say if we were to create an AI it would be for a better purpose than just cats and dogs… like a auto generating platformer or a AI that is pathfinding.
yes, that too.
You would need a minimum of 10,000 images to get a decent accuracy (60+%*) on the cross validation set. You will require a larger dataset to perform better. ( 60% is just a ballpark that we experienced , it may be better or worse for your dataset , you could establish a baseline using SVM one vs all strategy) .
https://www.quora.com/How-many-images-data-do-I-need-to-start-training-a-deep-neural-network-from-scratch-1?share=1
It's still less than the 200 GIGABYTES that are apparently needed.