Skip to content

Files

Latest commit

 

History

History
39 lines (29 loc) · 2.47 KB

README.md

File metadata and controls

39 lines (29 loc) · 2.47 KB

Program to detect object in a live stream

This project is a part of academic curriculum.

System Requirements:

  1. Laptop or a desktop with webcam set up. (Requires permission to use webcam. The project does not communicate anything remotely. All things happen only locally :) )

Instructions to run the program:

  1. Install python3 if not available in your system
  2. Install opencv-python package using pip installer.
  3. Go the program directory and run the following command: python3 main.py <lower_end_of_range> <higher_end_of_range>

Description about the command-line arguments:

main.py is the driver program that starts the app. lower_end_of_range is the lower end of the range of colors to be detected. higher_end_of_range is the higher end of the range of colors to be detected.

Why LOWER_END_OF_RANGE and UPPER_END_OF_RANGE?

The program begins by opening the webcam or any other primary camera of the system where it is run.

The user has to choose either l or r or q to either start recording a live stream or read a recorded video or quit the application respectively.

Live stream ( l ): On pressing c (for capture) key, the program captures a frame and does the following:

  1. It saves the captures frame (in RGB) format as an image file (rgb.jpg).
  2. It converts the RGB image to HSV format and saves it as a new image file (hsv.jpg).
  3. It then detects the object using the color range specified in the command-line arguments as lower_end_of_range and higher_end_of_range. This produces a black and white image (not a grayscale but a binary black/white image). The area where the object is present is in white and the other areas will be black.
  4. It saves the object-tracked image as a new image file (rgb-contoured.jpg).

On pressing Esc key with the focus on the live streaming window, the live stream recording stops.

Read a recorded video ( r ): On pressing c, the program captures a frame and does the following:

  1. The program then adds salt and pepper noise with density 0.02 to the original RGB image and saves it as a new image file (sp-noise-added.jpg).
  2. The program then applies the median filter algorithm to remove the salt and pepper noise and saves the result of the algorithm as a new image (sp-noise-removed.jpg).

On pressing Esc key with the focus on the video playback window, the video playback stops.

Note: The accuracy of the program depends on the range of the colors specified in the command line argument while running the program.