Read training: Handwritten digits recognition using Tensorflow with Python
I tried to move it to my ipynb.
There are some bugs in the training document, I fixed it and it can runs and output:
Model Execution Complete
But there is no information how to implement the model to recognize digits? The accuracy 0.2332 seems to be a bad result. How to improve it?
I search again and find a *better* tutorial in Chinese. It got accuracy around 0.9. The description is better.
The question so far is how to implement the model to recognize digits?
I’m a graphic designer, but I want to know machine learning. It is possible? So that I create this post to keep tracking my study path. It is an on going project, the post will be updated day by day.
Play face recognize
Using Python face_recognition to train machine recognize Produce 48 1AM group members. They are 34-An-YuJin, 52-Choi-YeNa, 56-Lee-KaEun, 57-Huh-YoonJin, 74-Takahashi-Juri.
Write a Python script to auto capture and crop faces from a video. There is the result:
Using Python face_recognition to train machine recognize Produce 48 1AM group members. …
I love design, I love computing.