pytorch_image_classification

Image Classification implemented in PyTorch. The dataset contains 28x28 (px) images of digits 0 to 8 in grayscale.Aim at recognizing the digit on the wooden block. This repository includes:

1. self-collected data set
2. training and testing the model
3. a config.jason for setting
4. the net refer to utils/net.py

Requirements

• torch
• torchvision
• numpy
• cv2
• PIL
• jason

Dataset

Train Model

All the parameters are in config.json

python train.py

Test Model

All the parameters are in config.json

python test.py or python infer.py