Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/timofurrer/hslu-pren-digit-cnn
Convolutional Neural Network to recognize digits used in the PREN class @ HSLU
https://github.com/timofurrer/hslu-pren-digit-cnn
cnn digit-recognition hslu keras machine-learning pren tensorflow
Last synced: about 1 month ago
JSON representation
Convolutional Neural Network to recognize digits used in the PREN class @ HSLU
- Host: GitHub
- URL: https://github.com/timofurrer/hslu-pren-digit-cnn
- Owner: timofurrer
- License: mit
- Created: 2019-07-03T17:48:25.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2019-07-03T22:09:46.000Z (over 5 years ago)
- Last Synced: 2024-10-30T03:57:05.419Z (3 months ago)
- Topics: cnn, digit-recognition, hslu, keras, machine-learning, pren, tensorflow
- Language: Jupyter Notebook
- Size: 7.78 MB
- Stars: 3
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- awesome-hslu - timofurrer/hslu-pren-digit-cnn - Data Augmentation & CNN for number detection (HSLU Modules / PREN2)
README
# CNN to detect digits for the PREN project @ HSLU
This repository contains notebooks, assets and a pre-trained model for a
Convolutional Neural Network intended to predict digits from 28x28 images.This CNN was used in the PREN project @ HSLU.
The repository consists of 3 parts:
* A jupyter notebook to augment and multiply the test images in the `images/` folder
* A jupyter notebook to train a CNN using the augmented test images
* A pre-trained model in `models/`The CNN classifies the numbers *1-9* and *0* for *not-a-number*.
See the jupyter notebooks for more details about the image augmentation and model architecture, training and evaluation.
Graph:
![](docs/tensorboard-logs.png)Accuracy during training:
![](docs/tensorboard-acc.png)***
*
This project is published under [MIT](LICENSE).
*
A [Timo Furrer](https://tuxtimo.me) project.
- :tada: -