Thai digit handwriting and example code Thai-MNIST is a dataset of thai handwrite digit images—consisting of a training set of 280,000 examples and a test set of 31,000 examples (75,000 examples and a test set of 8,400 examples in Alphabet Class) Each example is a 28x28 grayscale image, associated with a label from 10 classes. We intend Thai-MNIST to serve as a direct drop-in replacement for the original MNIST dataset for benchmarking machine learning algorithms. It shares the same image size and structure of training and testing splits.
-
Number dataset for train and validate download
-
Alphabet dataset for train and validate download
-
Number Image for train and validate download
-
Alphabet Image for train and validate download
- Python >= 3.4
- TensorFlow >= 1.0
- keras
- numpy
- cv2
- PIL
Download and Extract dataset to dataset/number
if you download alphabet class Extract dataset to dataset/alphabet
training model
python Thai_MNIST_Example.py
then testing model
python prediction_Thai_MNIST.py path/to/your/imge
We have a study section in jupyter notebook .
Thai_MNIST_Example.ipynb
We gets to 99.26% test accuracy after 10 epochs (Number class) and 99.58% (Alphabet class)
E-mail : chatchai@nextwaver.com
Tel : 0858242000
Line ID : 0858242000
E-mail : chatchai@nextwaver.com
Tel : 0858242000
Line ID : 0858242000
E-mail : chatchai@nextwaver.com
Tel : 0858242000