Traffic_Sign_Recognization

With the use of Kaggle Traffic Sign Dataset, A CNN model is trained to recognize traffic sign with more than 43 classes.

This project uses MLFLOW for logging purpose and contains github worflows aswell for ci-cd pipeline.
Project is deployed on Heroku.
To deploy the project on your heroku server configure
HEROKU_API_TOKEN: ${{ secrets.HEROKU_API_TOKEN }}
HEROKU_APP_NAME: ${{ secrets.HEROKU_APP_NAME }}
on your github account.

Data set Used

https://www.kaggle.com/meowmeowmeowmeowmeow/gtsrb-german-traffic-sign

WebApp Demo At

https://trafficsign-recognizer.herokuapp.com/

To run mlflow ui

Run Mlflow before running any training. This will setup mlflow db and experiment for us.
mlflow server --backend-store-uri sqlite:///mlflow.db --default-artifact-root ./mlruns

To Edit the model architecture

Edit src/utils/model_uti.py

To edit just model parameters

Edit config.yaml

Run Training after editing

python src/training.py
Each training will be logged to Mlflow with parameter,architecture,metrics and model

After some different model architectures and paramter, try to log best model to production.

To deploy a model into production based on some score

python src/log_production_model.py

To run the webapp

python app.py

Present Model Architecture

    Conv2D(filters=32,kernel_size=(5,5),activation='relu',input_shape=X_train.shape[1:]),
    Conv2D(filters=32,kernel_size=(5,5),activation='relu'),
    MaxPool2D(pool_size=(2,2)),
    Dropout(rate=0.2),
    Conv2D(filters=64,kernel_size=(3,3),activation='relu'),
    Conv2D(filters=64,kernel_size=(3,3),activation='relu'),
    MaxPool2D(pool_size=(2,2)),
    Dropout(rate=0.2),
    Flatten(),
    Dense(units=256,activation='relu'),
    Dropout(rate=0.4),
    Dense(units=43,activation='softmax')

With epochs=20, optimizer=adam and batch_size=32 got an accuracy of 0.95