https://github.com/breandan/comp551-p3
ConvNet training for our COMP 551 Project.
https://github.com/breandan/comp551-p3
Last synced: 4 months ago
JSON representation
ConvNet training for our COMP 551 Project.
- Host: GitHub
- URL: https://github.com/breandan/comp551-p3
- Owner: breandan
- Created: 2017-11-10T08:42:25.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2017-11-14T03:32:11.000Z (over 8 years ago)
- Last Synced: 2025-01-28T00:42:03.700Z (over 1 year ago)
- Language: Python
- Homepage: https://www.kaggle.com/c/comp551-modified-mnist
- Size: 2.68 MB
- Stars: 0
- Watchers: 4
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# COMP551 - Modified MNIST Challenge
## Prerequisites
Dependences:
```
Python 3.4+
h5py (pip)
opencv-python (pip)
TensorFlow 1.4+
```
To train the CNN, the following files should be present:
```
data/train_x.csv
data/train_y.csv
data/test_x.csv
```
These files can be retrieved from the COMP 551 [Kaggle competiton](https://www.kaggle.com/c/comp551-modified-mnist/data).
## Train the model
To train the model, simply run the following script: `python model_baseline.py`. This will save a model file to `data/temp_model.hdf5`.
To monitor training progress, run: `tensorboard --logdir=logs`
## Classify the data
To classify the test data, run: `python classify_data.py [optional_model_file.hdf5 (defaults to data/temp_model.hdf5)] [optional_output_file (defaults to data/test_y.csv)]`.