Object Detection Using TensorFlow

This repository contains implementation of object detection in live video-stream and also in an image using TensorFlow Object Detection API. Before running the code, you need to first install tensorflow and all of its dependencies.

System Requirements

* Ubuntu 18.04 LTS Bionic Beaver or Later  
* Intel Core i3 Processor  
* Dependencies Installed  

Note: Above requirements is according to my system.
Below installation procedure is for CPU not for GPU powered system.
For installation of TensorFlow on GPU, refer TensorFlow GPU Installation

Installation:

pip is a python package manager used for installing all the modules, if pip is not installed, then run the following command before insalling dependencies:

pip Installation:

For python 2.7:
   sudo apt-get -y install python-pip python-dev build-essential  
   sudo pip install --upgrade pip  
For python 3:  
   sudo apt-get -y install python3-pip  
   pip3 -version (For checking the version of pip3)

Dependencies Installation:

From the terminal run the following command:

    bash dependecies.sh

Note: dependencies.sh file contains all the dependencies/packages in order to run the program.

• In python interpreter, run the following commands to check whether TensorFlow is installed successfully: Open the terminal and write python3 or python depending upon your python version and run the following code:

import tensorflow as tf
print(tf.__version__) # print the version of tensorflow installed.

If you get the following output, then TensorFlow is installed successfully:

   '1.12.0' (Version of tensorflow installed on your system)

For detailed information on installation of TensorFlow, refer the below link:
TensorFlow Installation
TensorFlow Object Detection API Installation

OpenCV Installation:

• I used OpenCV for extracting frames from the camera in real-time.
• Refer the below link for installation OpenCV Installation