YOLO v8n - keep original IDs of objects, once other objects are not detected

I am using YOLOv8 with track in python. I can get the bounding boxes and classes of each object, but the IDs change every time one of the objects is no longer visible. If current object IDs are 1,2,3,4 and number 2 is not detected, YOLO simply decrements 3 and 4, so new IDs are 1,2,3, forcing me to manually detect which one left, which is a pain. I know YOLO can do this for me, as the show=True method for track plots the correct IDs even after objects are missed or leave the frame. I have tried looking in documentation but have not been able to find anything. The object returned by invoking track has methods sort, and index, which might be of help, but I don't know how to use them.

Small example:

from ultralytics import YOLO
model = YOLO("yolov8n-seg.pt")
results = model.track(img, conf=0.20, show = True, persist = True, tracker="botsort.yaml", verbose=False )
print(results[0].boxes.data[:,4])
----------------------------------
1,2,3,4
# when 2 is not detected:
1,2,3

Solution

check official documentation about persisting tracks loop.

You've got to use model.track frame by frame in loop with persist=True parameter. Furthermore be shure that you don't use stream=True parameter, it will breake your track's id.

so here is code from documentation with my correction about "finished tracks" (the objects that is no longer visible)

from collections import defaultdict

import cv2
import numpy as np

from ultralytics import YOLO

# Load the YOLOv8 model
model = YOLO('yolov8n.pt')

# Open the video file
video_path = "path/to/video.mp4"
cap = cv2.VideoCapture(video_path)

# Store the track history
track_history = defaultdict(lambda: [])

# Loop through the video frames
while cap.isOpened():
    # Read a frame from the video
    success, frame = cap.read()

    if success:
        # Run YOLOv8 tracking on the frame, persisting tracks between frames
        results = model.track(frame, persist=True)

        # Get the boxes and track IDs
        boxes = results[0].boxes.xywh.cpu()
        track_ids = results[0].boxes.id.int().cpu().tolist()

        # Visualize the results on the frame
        annotated_frame = results[0].plot()

        # Plot the tracks
        for box, track_id in zip(boxes, track_ids):
            x, y, w, h = box
            track = track_history[track_id]
            track.append((float(x), float(y)))  # x, y center point
            if len(track) > 30:  # retain 90 tracks for 90 frames
                track.pop(0)

            # Draw the tracking lines
            points = np.hstack(track).astype(np.int32).reshape((-1, 1, 2))
            cv2.polylines(annotated_frame, [points], isClosed=False, color=(230, 230, 230), thickness=10)

        # get finished tracks and do some logic with them
        finished_tracks = track_history.keys() - track_ids
        for ft_id in finished_tracks:
            ft = track_history.pop(ft_id)
            # do some logic with ft here.........

        # Display the annotated frame
        cv2.imshow("YOLOv8 Tracking", annotated_frame)

        # Break the loop if 'q' is pressed
        if cv2.waitKey(1) & 0xFF == ord("q"):
            break
    else:
        # Break the loop if the end of the video is reached
        break

# Release the video capture object and close the display window
cap.release()
cv2.destroyAllWindows()