image-classification-rnn - GitLite

KO | EN

gitlite — search

#python #react #ai #go #android #deep-learning #linux #nodejs #macos #awesome #awesome-list #security #ios #open-source

image-classification-rnn

(★ 90)

Classify MNIST image dataset into 10 classes. Build an image classifier with Recurrent Neural Network (RNN: LSTM) on Tensorflow.

File Explorer

Download Latest Version (.zip)

.gitignore
LICENSE
model.py
predict.py
README.md
requirements.txt
train.py
training_parameters.json

# Installation Guide

1. Get the code

git clone https://github.com/jiegzhan/image-classification-rnn

Downloads the entire project code from GitHub to your computer.

cd image-classification-rnn

Moves into the project folder you just downloaded.

2. Python

Easy Recommended

Prerequisites

Git Needed to download the project code from GitHub.
Python 3 On Windows, be sure to check 'Add Python to PATH' during installation.

python3 -m venv .venv

Runs the Python script (or module).

pip install -r requirements.txt

Installs the Python libraries listed in requirements.txt (or similar).

✅ If it runs without errors and prints output in the terminal, it worked.

Pulled directly from this repo's README.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

### Project: Build an Image Classifier with RNN (LSTM) on TensorFlow

### Highlights

- This is a **multi-class image classification** problem.
- The purpose of this project is to **classify the MNIST image dataset into 10 classes**.
- The model is built with a **Recurrent Neural Network (RNN: LSTM)** on **TensorFlow 2.x / Keras**.
- Each 28×28 image is treated as **28 timesteps × 28 features** (one row per timestep).

### Setup

```bash
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
```

### Data

- MNIST is downloaded automatically on first run via `tf.keras.datasets.mnist`.

### Train

```bash
python3 train.py ./training_parameters.json
```

A directory `trained_model_<timestamp>/` is created during training. The model is saved as `model.keras` inside that directory.

| Key | Default | Description |
|-----|---------|-------------|
| `learning_rate` | 0.001 | Adam learning rate |
| `training_iters` | 10000 | Approximate number of training samples to process |
| `batch_size` | 64 | Mini-batch size |
| `display_step` | 10 | Print progress every N epochs |

### Predict

Provide the model directory (created when running `train.py`) to `predict.py`:

```bash
python3 predict.py ./trained_model_<timestamp>/
```

### Reference

- [Recurrent Neural Network (TensorFlow-Examples)](https://github.com/aymericdamien/TensorFlow-Examples/blob/master/examples/3_NeuralNetworks/recurrent_network.py) — original pattern this repo modernized for TensorFlow 2.x.