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.
* 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
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:
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)
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 writepython3
orpython
depending upon yourpython
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
- I used
OpenCV
for extracting frames from the camera in real-time. - Refer the below link for installation OpenCV Installation