This is my code:

`from depthai_sdk import OakCamera
import depthai as dai

custom_model_path = 'model/CVAT2_blank_OISMRF_batch_16_ep_50_.json'
video_path = 'video/i8auditcam_2024-Jun-6_1207af80-840e-4525-b44b-f120c36d5cae_1920x1080.mp4'

with OakCamera(replay=video_path) as oak:
`` color = oak.create_camera('color')
nn = oak.create_nn(custom_model_path, color, tracker=True)

nn.config_nn(resize_mode='letterbox')
nn.config_tracker(
tracker_type=dai.TrackerType.SHORT_TERM_IMAGELESS,
track_labels=[0, 1, 2], # Track only specified object types
assignment_policy=dai.TrackerIdAssignmentPolicy.SMALLEST_ID,
max_obj=10, # Max objects to track
threshold=0.1 # Tracker threshold
)
oak.start()` ``

I want to save info such as trackid, bboxes, frameno, device_clocktime, label, confidence to a file. How do i achieve this

Hey This helped me. This is my code. Now i would like to save the bbox drawn video into a path.

    AmunTimilsina
    You can use openCV to draw the bbox over the tracker packet frame, then save it using cv.imwrite(path).

    Thanks
    Jaka

    Yah i did that. Also, you can see i save the results to a file.. i was analyzing the file and i got the same track id but different classes/Labels. Thats very weird. It happens for most of the tracks.

    @AmunTimilsina ,
    Seems like a problem with the accuracy of the detection model? Please also share some video of this, would help for debugging.