🚦
SignVision Lab
MobileNetV2 recognition pipeline
TensorFlow Traffic Sign Recognition
Train, evaluate, export, and predict traffic signs.
A complete VS Code runnable Python project using MobileNetV2, TensorFlow, ONNX export, automated dataset handling, and Streamlit inference.
25
Training epochs
224×224
Input resolution
Run commands
pip install -r requirements.txt
python src/train.py
streamlit run app.py
1
Download
Google Drive dataset fetched automatically with gdown
2
Prepare
Images resized to 224x224 and split into train validation test
3
Train
MobileNetV2 transfer learning with Adam and checkpoints
4
Evaluate
Accuracy precision recall F1 confusion matrix and accuracy plot
5
Export
TensorFlow SavedModel and ONNX model generated
6
Predict
Streamlit image upload returns traffic sign class