line counter is not working anymore?

[uzumakinaruto19]

Search before asking

• I have searched the Supervision issues and found no similar feature requests.

Question

this is my code I wanted to add multiple lines and count the objects, but the line counter is not working anymore with the latest supervision.

import os
import torch
os.environ["CUDA_DEVICE_ORDER"]="PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"]="0"
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
import numpy as np
import json
from ultralytics import YOLO
import supervision as sv
# from supervision.tools.line_counter import LineCounter
# from supervision.geometry.dataclasses import Point


# LINE_START = Point(50, 1500)
# LINE_END = Point(3790, 1500)

# line_counter = LineCounter(start=LINE_START, end=LINE_END)

model = YOLO('yolov8n.pt')

def main_func(input_data,frame):

    global model

    if 'ignore' in input_data:
        return "ignore"
    else:

        input_data = json.loads(input_data)

        results= model.track(frame, imgsz=640, classes=0,tracker="bytetrack.yaml")[0]
        detections = sv.Detections.from_yolov8(results)

        if results.boxes.id is not None:
            detections.tracker_id = results.boxes.id.cpu().numpy().astype(int)
            detections.conf = results.boxes.conf.cpu().numpy().astype(float)
            detections.xyxy = results.boxes.xyxy.cpu().numpy().astype(int)
            
            # print(detections.tracker_id)
            # print(detections.conf)
            # print(detections.xyxy)
        else:
            detections.tracker_id = np.array([])
            detections.conf = np.array([])
            detections.xyxy = np.array([])
 
        new_box=detections.xyxy
        new_box = new_box.astype("int").tolist()
        new_ids = detections.tracker_id.astype("int").tolist()
        new_confs = detections.conf.astype("float").tolist()    



        print("count",count)

        output = {
                  'frame_index':input_data['frame_index'],
                  'time_stamp':input_data['time_stamp'],
                  'detections':new_box,
                  'demo_run':input_data['demo_run'],
                  'ids':new_ids,
                  'confs':new_confs
                  }
        
        print(output)
        return output

except for the polygon , how to have multiple lines and update the counts?

Additional

No response

[github-actions]

Hello there, thank you for opening an Issue ! 🙏🏻 The team was notified and they will get back to you asap.

[SkalskiP]

Hi, @uzumakinaruto19 👋🏻! Could you be a bit more specific? What do you mean by "but the line counter is not working anymore with the latest supervision"?

[sceddd]

@uzumakinaruto19 did you use trigger() yet? I don't see it in your code.

[uzumakinaruto19]

Hi, @uzumakinaruto19 👋🏻! Could you be a bit more specific? What do you mean by "but the line counter is not working anymore with the latest supervision"?

hey sorry I'll provide the updated code - this my function for detection and tracking

def process_frame(input_data, frame, line_counter_1, line_counter_2, line_counter_3, line_annotator, box_annotator):

    
    results = model.track(frame, imgsz=640, classes=0, tracker="bytetrack.yaml")[0]
    detections = sv.Detections.from_yolov8(results)

    if results.boxes.id is not None:
        detections.tracker_id = results.boxes.id.cpu().numpy().astype(int)
        detections.conf = results.boxes.conf.cpu().numpy().astype(float)
        detections.xyxy = results.boxes.xyxy.cpu().numpy().astype(int)
    else:
        detections.tracker_id = np.array([])
        detections.conf = np.array([])
        detections.xyxy = np.array([])

    frame = box_annotator.annotate(scene=frame, detections=detections, labels=[str(id) for id in detections.tracker_id])
    line_counter_1.trigger(detections=detections)
    line_counter_2.trigger(detections=detections)
    line_counter_3.trigger(detections=detections)

    line_annotator.annotate(frame=frame, line_counter=line_counter_1)
    line_annotator.annotate(frame=frame, line_counter=line_counter_2)
    line_annotator.annotate(frame=frame, line_counter=line_counter_3)

    new_ids = detections.tracker_id.astype("int").tolist()
    new_box = detections.xyxy.astype("int").tolist()
    new_confs = detections.conf.astype("float").tolist()

    output = {
        'frame_index': input_data['frame_index'],
        'time_stamp': input_data['time_stamp'],
        'detections': new_box,
        'demo_run': input_data['demo_run'],
        'ids': new_ids,
        'confs': new_confs,
        'line_1_out_count': line_counter_1.out_count,
        'line_2_out_count': line_counter_2.out_count,
        'line_3_out_count': line_counter_3.out_count
    }

    print(output)
    return output




def display_frame(output, frame):
    cv2.imshow("Output", frame)
    cv2.waitKey(1)


# Provide the path to your video file
video_path = './test.mp4'
main_func(video_path)

even after using the trigger and the object crosses the line its not getting counted?

[uzumakinaruto19]

@uzumakinaruto19 did you use trigger() yet? I don't see it in your code.

i have used , I provided the wrong code, my bad

I think it had to do something with my tracking? well the IDs getting generated are not updating,
I have some issue with the tracker so the ID is getting resued every time, ie, when an object enters and that is assigned with an ID and and when that leaves the frame and a new object comes it is reusing that the ID
it should assign a new ID but it uses the number 1 again and start from there

i know tracking doesn't comes under this repo still , may that be an issue and how to tackle that, because the video might be an RTSP stream and id shouldn't be repeated just continuously and resued for at least sometime say like after 10 mins or something

@SkalskiP any help will be really appreciated

[SkalskiP]

Hi, @uzumakinaruto19 👋🏻

I'm not sure you can use YOLOv8 ByteTrack this way. Here is a snippet of code I used in the past, and it worked. Important note supervision API may differ a bit. It is pretty old code, but the YOLOv8 tracking part worked 100%.

import cv2

from ultralytics import YOLO
import supervision as sv
import numpy as np


LINE_START = sv.Point(320, 0)
LINE_END = sv.Point(320, 480)


def main():
    line_counter = sv.LineZone(start=LINE_START, end=LINE_END)
    line_annotator = sv.LineZoneAnnotator(thickness=2, text_thickness=1, text_scale=0.5)
    box_annotator = sv.BoxAnnotator(
        thickness=2,
        text_thickness=1,
        text_scale=0.5
    )

    model = YOLO("yolov8l.pt")
    for result in model.track(source=0, show=True, stream=True, agnostic_nms=True):
        
        frame = result.orig_img
        detections = sv.Detections.from_yolov8(result)

        if result.boxes.id is not None:
            detections.tracker_id = result.boxes.id.cpu().numpy().astype(int)
        
        detections = detections[(detections.class_id != 60) & (detections.class_id != 0)]

        labels = [
            f"{tracker_id} {model.model.names[class_id]} {confidence:0.2f}"
            for _, confidence, class_id, tracker_id
            in detections
        ]

        frame = box_annotator.annotate(
            scene=frame, 
            detections=detections,
            labels=labels
        )

        line_counter.trigger(detections=detections)
        line_annotator.annotate(frame=frame, line_counter=line_counter)

        cv2.imshow("yolov8", frame)

        if (cv2.waitKey(30) == 27):
            break


if __name__ == "__main__":
    main()

[uzumakinaruto19]

Hi, @uzumakinaruto19 👋🏻

I'm not sure you can use YOLOv8 ByteTrack this way. Here is a snippet of code I used in the past, and it worked. Important note supervision API may differ a bit. It is pretty old code, but the YOLOv8 tracking part worked 100%.

import cv2

from ultralytics import YOLO
import supervision as sv
import numpy as np


LINE_START = sv.Point(320, 0)
LINE_END = sv.Point(320, 480)


def main():
    line_counter = sv.LineZone(start=LINE_START, end=LINE_END)
    line_annotator = sv.LineZoneAnnotator(thickness=2, text_thickness=1, text_scale=0.5)
    box_annotator = sv.BoxAnnotator(
        thickness=2,
        text_thickness=1,
        text_scale=0.5
    )

    model = YOLO("yolov8l.pt")
    for result in model.track(source=0, show=True, stream=True, agnostic_nms=True):
        
        frame = result.orig_img
        detections = sv.Detections.from_yolov8(result)

        if result.boxes.id is not None:
            detections.tracker_id = result.boxes.id.cpu().numpy().astype(int)
        
        detections = detections[(detections.class_id != 60) & (detections.class_id != 0)]

        labels = [
            f"{tracker_id} {model.model.names[class_id]} {confidence:0.2f}"
            for _, confidence, class_id, tracker_id
            in detections
        ]

        frame = box_annotator.annotate(
            scene=frame, 
            detections=detections,
            labels=labels
        )

        line_counter.trigger(detections=detections)
        line_annotator.annotate(frame=frame, line_counter=line_counter)

        cv2.imshow("yolov8", frame)

        if (cv2.waitKey(30) == 27):
            break


if __name__ == "__main__":
    main()

@SkalskiP thanks, I've referred while doing it

well the detection is working fine aswell as the tracking

below shown is the output

WARNING ⚠️ imgsz=[740] must be multiple of max stride 32, updating to [768] {'frame_index': 10, 'time_stamp': 0, 'detections': [[275, 147, 317, 271], [330, 162, 347, 221]], 'demo_run': False, 'ids': [1, 2], 'confs': [0.8728579878807068, 0.731636643409729], 'line_1_out_count': 0} Line 1 Out Count: 0

I'm having issues while counting alone ids what I'm guessing

I have drawn the line but even after the object crossing it is not getting counted

[SkalskiP]

With the current implementation, for the object to be counted, all four corners of the bounding box must cross the line.

[uzumakinaruto19]

So u are suggesting extending the line more (I guess I tried that aswell, it is not getting counted at all, anyway I'll try once more), should a go with a polygon(idk whether that will make any sense with that)

[SkalskiP]

I actually suggest putting the camera further away.

[uzumakinaruto19]

With the current implementation, for the object to be counted, all four corners of the bounding box must cross the line.

Instead of that can't u use the centroid of the Box, more accurate maybe?

[SkalskiP]

Using single point with models like YOLOv8 is not reliable solution unfortunately.

[sceddd]

Maybe you put the line in the wrong position. The logic is that if all the bounding boxes cross the line, the count will increase. In your case, you should either turn the camera to a higher position or detect the shoes and count if those shoes cross the line or you could take a look at #107 with the @maddust issue by choose the point at bottom center of the bounding box for counting.