Skip to content

Latest commit

 

History

History
56 lines (49 loc) · 2.27 KB

README.md

File metadata and controls

56 lines (49 loc) · 2.27 KB

5(-2)K - Social distancing, face masks, and no gatherings

Covid-negative but positive in loving you <3.

Abstract

Pipeline

Workflow

Installation

  1. Require Python 3.6+
  2. Clone this repository
  3. Go to src folder and run
    pip install -r requirements.txt

Usage

  1. Go to src/distance folder and run

    python distance.py --video path

    First frame will be shown, then you have to click to choose 6 points:

    • First 4 points are the top-left, top-right, bottom-left, bottom-right corners of the ROI.
    • The last 2 points define distance in horizontal and vertical, which equals to 1 meter in the real world.

    The result will to save to distance.txt.

  2. Go to src folder and download pretrained weights for YOLOv5, make sure pt files in weights folder.

  3. Go to src folder and run

    bash demo.sh <video path> <limit number of people>

    The result will be saved in src\runs\track\test\<video name>

Demo

Demo

Demo link:

Google Colab demo
Open In Colab

Acknoledgements

First, cam on giang vien.

Second, allmost our code is borrowed from these resources:

  1. YOLOv5 and deepakcrk's pretrained weights for head & person detection.
  2. ByteTrack for tracking method.

Many thanks for their wonderful work.

Finally, team spirit is the most important factor to win victory. Thank you for all your hảd wỏk. We did very well :>.

Team members

On-going work

  • Person and head detection
  • Add tracking method to improve detection performance
  • Face mask classifier
  • Top-view localization and distance estimation
  • Optimize the face mask classifier
  • Train more yolov5 models at more scale