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.
* 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.
* 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.
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:
pip install git+https://github.com/swetha4444/PyTrackX.git
You can also clone the repository:
git clone https://github.com/swetha4444/PyTrackX.git
pip install -r requirements.txt
python setup.py install
* numpy>=1.11
* matplotlib>=1.5
* pandas
* opencv-python
* scipy
* mediapipe
from PyTrackX import *
yolo_path = "./yolo-coco"
video_path = "input.mp4"
object = "person"
track_object(yolo_path,video_path,object)
from PyTrackX import *
track_posture()