https://github.com/programmer-rd-ai/digivis
A PyTorch-based deep learning implementation for MNIST digit recognition featuring CNNs, GPU acceleration, experiment tracking, and comprehensive testing capabilities.
https://github.com/programmer-rd-ai/digivis
cnn computer-vision cuda data-science deep-learning digit-recognition image-classification machine-learning mnist neural-networks python pytorch wandb
Last synced: 14 days ago
JSON representation
A PyTorch-based deep learning implementation for MNIST digit recognition featuring CNNs, GPU acceleration, experiment tracking, and comprehensive testing capabilities.
- Host: GitHub
- URL: https://github.com/programmer-rd-ai/digivis
- Owner: Programmer-RD-AI
- License: apache-2.0
- Created: 2023-07-09T02:40:28.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2025-01-12T11:17:45.000Z (5 months ago)
- Last Synced: 2025-03-01T23:52:19.285Z (4 months ago)
- Topics: cnn, computer-vision, cuda, data-science, deep-learning, digit-recognition, image-classification, machine-learning, mnist, neural-networks, python, pytorch, wandb
- Language: Jupyter Notebook
- Homepage: https://wandb.ai/ranuga-d/DigiVis/
- Size: 82 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 14
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# DigiVis: Neural Network Vision for Digit Classification
A deep learning implementation for MNIST digit recognition using convolutional neural networks and computer vision techniques. This project combines modern neural architectures with advanced image processing for accurate digit classification.
## Overview
DigiVis is a comprehensive implementation of various neural network architectures for digit recognition, utilizing the MNIST dataset. The project incorporates modern deep learning practices including data normalization, image transformations, and model evaluation metrics.
## Features
- Multiple neural network architectures (CNN and Linear models)
- Data normalization and preprocessing
- Image transformations and augmentations
- Model training with performance metrics
- Weights & Biases integration for experiment tracking
- Comprehensive test suite
- CUDA support for GPU acceleration
## Requirements
- Python 3.x
- PyTorch
- torchvision
- numpy
- pandas
- Pillow
- wandb
- matplotlib
- scikit-learn
- tqdm
## Installation
1. Clone the repository
```bash
git clone https://github.com/Programmer-RD-AI/DigiVis.git
```
2. Install dependencies
```bash
pip install -r requirements.txt
```
## Usage
Run the main training script:
```
python run.py
```
For interactive exploration, use the provided Jupyter notebook:
```
jupyter notebook test.ipynb
```
## Model Configuration
- Image Size: 224x224
- Batch Size: 32
- CUDA enabled for GPU acceleration
- Random seed: 42 for reproducibility
## License
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.