maloleg / simple-neuro

trying to make neural network which differs triangles, squares, circles

Simple-neuro

trying to make neural network which differs triangles, squares, circles
Tensorflow 2.4
MakingDataSet.py - literally making MNIST like data set, which contains 150 pictures of triangles, squares, circles
actualTesting.py - creating model, learning, testing with pictures in tests directory
./triangles, ./squares ./circles - folders with raw pictures (primarily from https://yandex.ru/images/ due to possibility of format filter) and figure/done folders with 50x50 greyscale images
overall result: not so bad with other pictures from the internet and almost none result with handwritten images