jalajthanaki / Real_time_object_detection

This repository contains the code for real-time object detection. I'm using video stream coming from webcam. MobileNet-SSD and OpenCv has been used as base-line approach. TensorFlow object detection API has been used in revised approach.

Real time object detection

Overview

This repository contains the code for real time object detection.

There is two approach given here.

1. Base line or basic approach
2. revised approach

Dependencies and installation

For base line approach there are following libraries which are needed.

• imutils: $ sudo pip install imutils
• Install computervision or CV2 from here
• time - default library of python
• numpy - $ sudo pip install numpy

For revised approach there are following libraries which are needed.

• os - default library of python
• Install computervision or CV2 from here
• time - default library of python
• argparse - default library of python
• multiprocessing - default library of python
• numpy $ sudo pip install numpy
• tensorflow - You can refer this link

Usage

Run base line approach using following execution steps.

1. Go to the base_line_model folder.
    $ cd ./base_line_model
2. To run the script execute following command.
    $ python object_detection_app.py

Run revised approach using following execution steps.

1. Go to the revised_approach folder.
    $ cd ./revised_approach
2. To run the script execute following command.
    $ python real_time_object_detection.py --prototxt MobileNetSSD_deploy.prototxt.txt --model MobileNetSSD_deploy.caffemodel 

Credit

The credit of this code goes to Dat Tran and the blogs provided on Pyimagesearch site. I've merely created a wrapper to get people started.