Deep Learning Traffic Sign Recognition System
A Deep Learning Application for Traffic Sign Recognition
- Python 3.6+
- Ubuntu 16.04LTS
- Virtual Environment for Python with pip
- Anaconda
virtualenv
,pyvenv
- Network Access for getting dependencies
- Clone Repo
git clone https://github.com/rachit-ranjan16/Kiera.git
- Activate Virtual Environment
- Install RabbitMQ
sudo apt-get -y -q install rabbitmq-server
- Get project dependencies in place
cd Kiera
python setup.py develop
- Start up Django(Dev Mode)
python manage.py runserver 0.0.0.0:8080
- Start up Celery(Dev Mode)
celery -A Kiera worker -l info
- Layer 1 Input Conv
- kernel 5x5
- stride 1x1
- reLU activation
- Layer 2 Conv
- kernel 5x5
- stride 1x1
- reLU activation
- Layer 3 MaxPooling
- pooling size 2x2
- dropout 0.25
- Layer 4 Conv
- kernel 5x5
- stride 1x1
- reLU activation
- Layer 5 Conv
- kernel 5x5
- stride 1x1
- reLU activation
- Layer 6 MaxPooling
- pooling size 2x2
- dropout 0.25
- Layer 7 Flatten
- Layer 8 Fully Connected
- dropout 0.25
- relU activation
- Layer 9 Output
- SoftMax Classification