infer_boxmot_tracking

Multiple object tracking algorithm using BoxMOT trackers for object detection, instance segmentation and keypoints.

BoxMOT provides several tracking algorithms such as OccluBoost, BoT-SORT, BoostTrack, StrongSORT, DeepOCSORT, ByteTrack, HybridSORT, OC-SORT and SFSORT.

🚀 Use with Ikomia API

1. Install Ikomia API

We strongly recommend using a virtual environment. If you're not sure where to start, we offer a tutorial here.

pip install ikomia

2. Create your workflow

from ikomia.dataprocess.workflow import Workflow
from ikomia.utils.displayIO import display
import cv2

# Init your workflow
wf = Workflow()

# Add object detection algorithm
detector = wf.add_task(name="infer_yolo_26", auto_connect=True)

# Add BoxMOT tracking algorithm
tracking = wf.add_task(name="infer_boxmot_tracking", auto_connect=True)

stream = cv2.VideoCapture(0)
while True:
    # Read image from stream
    ret, frame = stream.read()

    # Test if streaming is OK
    if not ret:
        continue

    # Run the workflow on current frame
    wf.run_on(array=frame)

    # Get results
    image_out = tracking.get_output(0)
    obj_track_out = tracking.get_output(1)

    # Display
    img_res = cv2.cvtColor(image_out.get_image_with_graphics(obj_track_out), cv2.COLOR_BGR2RGB)
    display(img_res, title="BoxMOT tracking", viewer="opencv")

    # Press 'q' to quit the streaming process
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# After the loop release the stream object
stream.release()
# Destroy all windows
cv2.destroyAllWindows()

☀️ Use with Ikomia Studio

Ikomia Studio offers a friendly UI with the same features as the API.

• If you haven't started using Ikomia Studio yet, download and install it from this page.
• For additional guidance on getting started with Ikomia Studio, check out this blog post.

📝 Set algorithm parameters

• tracker (str) - Default 'occluboost': Tracking algorithm to use. Available values are 'occluboost', 'botsort', 'boosttrack', 'strongsort', 'deepocsort', 'bytetrack', 'hybridsort', 'ocsort' and 'sfsort'.
• reid (str) - Default 'osnet_x0_25_msmt17': Re-identification model used by trackers that support ReID.
• config (str) - Default '': Optional path to a custom tracker configuration file in YAML format.
• categories (str) - Default 'all': Categories of objects you want to track. Use a comma separated string to set multiple categories (ex: "dog,person,car").
• cuda (bool) - Default depends on CUDA availability: Enable CUDA acceleration when available.
• half (bool) - Default depends on CUDA availability: Enable half precision when CUDA is available.

Available ReID models in BoxMOT 21.0.0:

• resnet50_market1501
• resnet50_dukemtmcreid
• resnet50_msmt17
• resnet50_fc512_market1501
• resnet50_fc512_dukemtmcreid
• resnet50_fc512_msmt17
• mlfn_market1501
• mlfn_dukemtmcreid
• mlfn_msmt17
• hacnn_market1501
• hacnn_dukemtmcreid
• hacnn_msmt17
• mobilenetv2_x1_0_market1501
• mobilenetv2_x1_0_dukemtmcreid
• mobilenetv2_x1_0_msmt17
• mobilenetv2_x1_4_market1501
• mobilenetv2_x1_4_dukemtmcreid
• mobilenetv2_x1_4_msmt17
• osnet_x1_0_market1501
• osnet_x1_0_dukemtmcreid
• osnet_x1_0_msmt17
• osnet_x0_75_market1501
• osnet_x0_75_dukemtmcreid
• osnet_x0_75_msmt17
• osnet_x0_5_market1501
• osnet_x0_5_dukemtmcreid
• osnet_x0_5_msmt17
• osnet_x0_25_market1501
• osnet_x0_25_dukemtmcreid
• osnet_x0_25_msmt17
• osnet_ibn_x1_0_msmt17
• osnet_ain_x1_0_msmt17
• lmbn_n_duke
• lmbn_n_market
• lmbn_n_cuhk03_d
• clip_market1501
• clip_duke
• clip_veri
• clip_vehicleid

This list is read from BoxMOT at runtime and may change with future BoxMOT releases. Check the upstream mikel-brostrom/boxmot repository for the latest available models.

from ikomia.dataprocess.workflow import Workflow

# Init your workflow
wf = Workflow()

# Add detection algorithm
detector = wf.add_task(name="infer_yolo_26", auto_connect=True)

# Add tracking algorithm
tracking = wf.add_task(name="infer_boxmot_tracking", auto_connect=True)

tracking.set_parameters({
    "tracker": "bytetrack",
    "categories": "person,car",
    "cuda": "True",
    "half": "True",
})

🔍 Explore algorithm outputs

Every algorithm produces specific outputs, yet they can be explored them the same way using the Ikomia API. For a more in-depth understanding of managing algorithm outputs, please refer to the documentation.

from ikomia.dataprocess.workflow import Workflow

# Init your workflow
wf = Workflow()

# Add detection algorithm
detector = wf.add_task(name="infer_yolo_26", auto_connect=True)

# Add tracking algorithm
tracking = wf.add_task(name="infer_boxmot_tracking", auto_connect=True)

# Run on your image
wf.run_on(url="https://raw.githubusercontent.com/Ikomia-dev/notebooks/main/examples/img/img_work.jpg")

# Iterate over outputs
for output in tracking.get_outputs():
    # Print information
    print(output)
    # Export it to JSON
    output.to_json()

⏩ Advanced usage