Skip to content

swetha4444/PyTrackX

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

47 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation


A general python framework for visual object tracking. PyTrackX is a Python package that serves as an automated tool which tracks different objects and body movements as per user requirement and returns the real-time coordinates with very less steps and a simple video input.

Functions

track_object(yolo_path,video_path,object,min_conf)

* yolo_path: folder path for the download yolo model files (STRING)
* video_path: file path of video input feed
* object: any object name from list of objects YOLO can detect. 
* min_conf (default parameter): confidence level for an object detected (default value is 0.5)

Returns a text file with coordinates of each object detected per frame as tracked frame to frame. image

track_posture(width,height,video_path,min_dconf,min_tconf)

* width (default parameter): frame width
* height (default parameter): frame height
* video_path (default parameter): file path of video input feed (default value (0)- Live webcam video feed as input)
* min_dconf: confidence level for an object detected
* min_tconf: confidence level for an object tracking

Returns a text file with corrdinates of all landmarks detected by Mediapipe per frame as tracked frame to frame with respect to the frame width and height.
image

User Installation

Install Package

pip install PytrackX==0.7

Download yolo-coco folder from here into your current working directory. Rename the folder yolo-coco if otherwise. The folder contains the following files:

* coco.name
* yolo3.cfg
* yolo3.weights

Another option is to download via git:

Git Install

pip install git+https://github.com/swetha4444/PyTrackX.git

You can also clone the repository:

Clone Repository

git clone https://github.com/swetha4444/PyTrackX.git
pip install -r requirements.txt
python setup.py install

Dependencies:

* numpy>=1.11
* matplotlib>=1.5
* pandas
* opencv-python
* scipy
* mediapipe

Usage

YOLO Object Tracking

from PyTrackX import *
yolo_path = "./yolo-coco"
video_path = "input.mp4"
object = "person"
track_object(yolo_path,video_path,object)

Mediapipe Posture Tracking

from PyTrackX import *
track_posture()

Author

Sriram Kannan
Swetha Saseendran