LineZone detection failed with multiple line

[sceddd]

i found that in LineZone detection, you try to trigger with all the detections allready track by ByteTrack, but it failed when work with multible LineZone base on the logic that check all the tracker one and count them at the end of for loop.

[sceddd]

Hello, its me again.I have check your code again and figure out the line zone you use to count object will, also count the object outside the line, the object that don't even come close to the line. Here is the example base on the one on the tutorial video:

and also figure out that if you use the line that have the same direction with the one you already initialize, those in_count and out_count will return the same.

Hope you understand, im try to explain with my spoiled english.

[maddust]

Hey @sceddd there is an issue open that is related to this bug see issue #87 . I have solved the issue by modifying the logic a bit to check if the anchor is within the line segment. I Will update the one I shared in issue #87 since I had to add a "grace period" to the code to avoid double counting when a bounding box jumps.

[maddust]

@sceddd
here you go @SkalskiP will most likely will come up with a better solution but for now this works for me .

Adjust the grace period if you want more frames to be ignored until the object can be counted again
def init(self,name, start: Point, end: Point, grace_period: int = 30):

Adjust margin: threshold for the trigger to count the object
def is_within_line_segment(self, point: Point, margin: float = 5) -> bool:

from typing import Dict

import cv2
import numpy as np

from supervision.detection.core import Detections
from supervision.draw.color import Color
from supervision.geometry.core import Point, Rect, Vector
from typing import Optional

class LineZone:
def init(self,name, start: Point, end: Point, grace_period: int = 30):
self.name = name
self.vector = Vector(start=start, end=end)
self.tracker_state: Dict[str, bool] = {}
self.already_counted: Dict[str, bool] = {} # Flag for each tracker_id
self.frame_last_seen: Dict[str, int] = {} # Last frame each tracker_id was encountered
self.current_frame: int = 0 # Current frame counter
self.in_count: int = 0
self.out_count: int = 0
self.grace_period = grace_period # Number of frames to wait before resetting
def is_within_line_segment(self, point: Point, margin: float = 5) -> bool:
"""
Check if a point is within the line segment.

    Attributes:
        point (Point): The point to be checked.
        margin (float): Additional margin added to the line segment boundaries.

    Returns:
        bool: True if the point is within the line segment, False otherwise.
    """
    x_within = min(self.vector.start.x, self.vector.end.x) - margin <= point.x <= max(self.vector.start.x, self.vector.end.x) + margin
    y_within = min(self.vector.start.y, self.vector.end.y) - margin <= point.y <= max(self.vector.start.y, self.vector.end.y) + margin

    return x_within and y_within


def trigger(self, detections: Detections):
    self.current_frame += 1  # Increment frame counter

    for xyxy, _, confidence, class_id, tracker_id in detections:
        # handle detections with no tracker_id
        if tracker_id is None:
            continue

        # we check if the bottom center anchor of bbox is on the same side of vector
        x1, y1, x2, y2 = xyxy
        anchor = Point(x=(x1+x2)/2, y=y2)  # Bottom center point of bounding box

        # Check if anchor is within the line segment
        if not self.is_within_line_segment(anchor):
            continue

        tracker_state = self.vector.is_in(point=anchor)

        # handle new detection
        if tracker_id not in self.tracker_state:
            self.tracker_state[tracker_id] = tracker_state
            self.already_counted[tracker_id] = False  # When the object appears, it has not been counted yet
            self.frame_last_seen[tracker_id] = self.current_frame  # Update last seen frame
            continue

        # handle detection on the same side of the line
        if self.tracker_state.get(tracker_id) == tracker_state:
            continue

        # check if this crossing has already been counted
        if self.already_counted.get(tracker_id, False):  # If it has been counted, we skip this
            # Check if grace period has passed since last encounter
            if self.current_frame - self.frame_last_seen.get(tracker_id, 0) > self.grace_period:
                self.already_counted[tracker_id] = False  # Reset after grace period
            else:
                continue

        self.tracker_state[tracker_id] = tracker_state
        self.already_counted[tracker_id] = True  # After counting, we mark this crossing as already counted
        self.frame_last_seen[tracker_id] = self.current_frame  # Update last seen frame

        if tracker_state:
            self.in_count += 1
        else:
            self.out_count += 1

[sceddd]

Thank you @maddust, i'm very appreciate it.
One more question, in object tracking by ByteTracker, i realize that in a crowd or in a traffic jam, the tracking will not working well on it, like if there is a car hidding behind another one, it lead to a problem that the tracker id will be different if the detections.detect it once again. If i'd like to count the objects in the frame, it'll lose bunch of them. Any idea?

[maddust]

Thank you @maddust, i'm very appreciate it. One more question, in object tracking by ByteTracker, i realize that in a crowd or in a traffic jam, the tracking will not working well on it, like if there is a car hidding behind another one, it lead to a problem that the tracker id will be different if the detections.detect it once again. If i'd like to count the objects in the frame, it'll lose bunch of them. Any idea?

that is very subjective of how good your trained model is, if you are working with a default model like yolov8 or others ,they might not be good enough for your particular situation(you will have to train a custom model) also you can play with the threshold settings of the tracker. the other thing i can suggest is to avoid setting a line or zones in parts where occlusion happens . i regularly counts traffic and crows of people . for traffic i set lines or zones where there is free flow of traffic . here is some example of my implementation of supervision and yolov8 https://youtu.be/dER7_RMAkuU hope it helps

[SkalskiP]

Hi, @maddust 👋🏻! Thanks a lot for taking the initiative and helping @sceddd. It makes me super happy that we have people like you - active and keen to help others - in the supervision community. 💜

As for your solution, it is a lot simpler than the one I proposed. I'm not sure if it will work in all cases. But @maddust and @sceddd feel free to use it until I'll come up with a more robust fix.