Pothole-Detection-Flask-API

A flask API that predicts if an image has a plothole in it using tensorflow model generated though CNN.

Installation

1. Install Anaconda.

Following instructions are given for anaconda installation in Ubuntu. Installation can also be done without anaconda using pipenv.

2. Create a new virtual environment.

tf is the name of environment.

conda create --name tf or conda create --name tf python=3.7 specify the version of python 3 you have

Activate the environment before installing libraries

conda activate tf

3. Install packages

Libraries required for running are as follows. Tensorflow 2.0, Keras 2.3, Opencv 3.4.1, numpy, matplotlib and Flask Install the libraries using conda by the following commands. Ensure the version of package before hitting y.

conda install numpy
conda install matplotlib
conda install tensorflow
conda install keras
conda install opencv

And install flask conda install flask

Running

Clone the repository

Activate the environment. conda activate tf Close any app taking too much memory before running to prevent unresponsive pc. It consumes a ram of about 6-7 GB. Having swap m

conda activate myenv
export FLASK_APP=app.py
flask run