Tracker

This package is an collection of three type of tracker. TRACKER_MAP = {'bytetrack': BYTETracker, 'botsort': BoTSORT, 'sort': Sort}

Example

Tracker class can be used to track the objects in a video.

Please check the [tracker.py](tracker/tracker.py) for more details.

# import tracker
from tracker.tracker import Tracker

# Create an instance of tracker
tracker = Tracker(tracker_type = 'sort')

# detections result from object detector
dets = np.array([[0,0,10,10,0.9,1],[0,0,10,10,0.8,1],[0,0,10,10,0.7,1], ....])

# update tracker
tracking_results = tracker(dets)

# print tracking results
print(tracking_results)

Class Explanation

• Tracker: This class will initialize the tracker with the given tracker_type and tracker parameters from corresponding yaml file. call will take the detections and return the tracking results. Please check the tracker.py for more details.

  • arguments:

    detections parameters in the format

          if tracker_type == 'bytetrack' or tracker_type == 'botsort':
              detections: Result class with following attributes
                  class Result:
                      def __init__(self) -> None:
                          self.xyxy = None
                          self.conf = None
                          self.cls = None
              img: np.ndarray image
          
          if tracker_type == 'sort':
              detections: new detections in the format
              detections: `[[x1, y1, x2, y2, score, class_id], ...]`
    

  • returns:

    • Returns the updated bounding boxes in the format [[x1, y1, x2, y2, track_id, score, class_id], ...] if tracker is not initalized it will return None

Use in your own peoject

Below is the gist of how to instantiate and update the tracker. Please check the tracker.py for more details.

1. using Sort Tracker

from tracker.tracker import Tracker
# Create an instance of tracker
obj_tracker = Tracker(tracker_type = 'sort')
# get detections
...

# update SORT
track_bbs_conf_cls_ids = obj_tracker(detections)

# track_bbs_conf_cls_ids is a np array where each row contains a valid bounding box, track_id, score and class
...

2. Using ByteTracker and Boat Tracker using yolov5 object detector

from tracker.tracker import Tracker
# Create an instance of tracker
obj_tracker = Tracker(tracker_type = 'bytetrack')

# create class to store results of each frame
class Result:
    def __init__(self) -> None:
        self.xyxy = None
        self.conf = None
        self.cls = None
# get detections
...
...

if len(det):
    # Rescale boxes from img_size to im0 size
    det[:, :4] = scale_boxes(im.shape[2:], det[:, :4], im0.shape).round()
    # detect results
    det1 = det.cpu().numpy()
    # update result class
    results.xyxy = det1[:, :4]
    results.conf = det1[:, 4]
    results.cls = det1[:, 5]
    # update Tracker
    det_S = obj_tracker(results, im0)
    if (det_S is not None) and (det_S.size != 0):
        for *xyxy, ID, conf, cls in reversed(det_S):
            # *xyxy : Bounding box
            # ID: id of each object
            # conf: confidence score
            # cls: class index
        ...
        ...
    ...
...

Build setup

# Go to code directory
cd some_root_dir

# create build

python -m build
# or
# python setup.py sdist bdist_wheel

It will create build in your dist directory. if dist is not present it will create it.

Install Packages

if you have not made any build, please follow the command

python setup.py install

If you have created an build, please follow the instruction below

cd dist
pip install package_name*.whl

Please replace "package_name*.whl" with .whl file present in dist directory

Work from source code

Clone repo and install requirements.txt

git https://github.com/amish0/Tracker.git
cd Tracker
pip install -r requirements.txt

References

- SORT: https://github.com/abewley/sort

- Byte Track: https://github.com/ifzhang/ByteTrack

- BoT-SORT: https://github.com/NirAharon/BoT-SORT