ID Not Writing To Label File

[treyrachall10]

Is there a reason why the ID sometimes will not write to the .txt file? For example frame n might have (0, 0.561567, 0.247865, 0.396077, 0.261982, 3) but frame n + 1 will have (0, 0.943255, 0.225126, 0.110261, 0.0804192)

[mikel-brostrom]

Which file? Could you give more context on what you are running?

[treyrachall10]

Yes I am running Track with DeepSORT. The problem is some of the label files it creates only have 5 columns and some have 6.

[mikel-brostrom]

DeepSORT does not exist in this repo

[treyrachall10]

I misspoke it's the deepocsort tracker that I am running.

[mikel-brostrom]

The benchmark pipeline passes without issues. Where MOT results are generated and fed to TrackEval for evaluation. I have never encountered this issue. Please provide more context of what you are running.

[treyrachall10]

For sure, maybe I am doing something wrong. Here is the exact command I am running "boxmot track --yolo-model yolov8s.pt" --reid-model "osnet_x1_0_msmt17.pt" --source "videopath" --tracking-method deepocsort --save-txt --verbose --classes 0". Correct me if this looks wrong, this is my first time working with detection and tracking. I am expecting to get an output of 6 columns in my .txt files in the label folder. I would say about 60% of them only have 5 which contain the class, bbox, bboxy, width, and height. But the others have class, bbox, bboxy, width, height, and ID. Attached are two examples from the same run. Thanks for trying to help me out.
bjjvideo_1.txt
bjjvideo_119.txt

[mikel-brostrom]

Aha, I see the issue now. Track does not support this generation at the moment. Try

$ boxmot eval --yolo-model yolov8n.pt --reid-model osnet_x0_25_msmt17.pt --tracking-method deepocsort  --source ./assets/MOT17-mini/train --verbose

but this will only work if you have a mot17/20/dancetrack-like folder structure

[treyrachall10]

Thank you for this! I will try this when I sit down to do some work later today!

[mikel-brostrom]

An easy way of enabling this is by mokey-patching ultralytics' save_txt, this will work:

# Mikel Broström 🔥 Yolo Tracking 🧾 AGPL-3.0 license

import argparse
from functools import partial
from pathlib import Path

import cv2
import torch

from boxmot import TRACKERS
from boxmot.tracker_zoo import create_tracker
from boxmot.utils import ROOT, TRACKER_CONFIGS, WEIGHTS
from boxmot.utils.checks import RequirementsChecker
from boxmot.engine.detectors import default_imgsz, get_yolo_inferer, is_ultralytics_model

checker = RequirementsChecker()
checker.check_packages(("ultralytics", ))  # install

from ultralytics import YOLO
from ultralytics.utils.plotting import Annotator  # ultralytics.yolo.utils.plotting is deprecated
from ultralytics.utils.plotting import colors
from ultralytics.utils import plotting

# Make every drawing call a no-op
plotting.Annotator.box       = lambda *args, **kwargs: None
plotting.Annotator.box_label = lambda *args, **kwargs: None
plotting.Annotator.line      = lambda *args, **kwargs: None


from pathlib import Path
from ultralytics.engine.results import Results

def save_txt(self, txt_file: Union[str, Path], save_conf: bool = False) -> str:
        """
        Save detection results to a text file.

        Args:
            txt_file (str | Path): Path to the output text file.
            save_conf (bool): Whether to include confidence scores in the output.

        Returns:
            (str): Path to the saved text file.

        Examples:
            >>> from ultralytics import YOLO
            >>> model = YOLO("yolo11n.pt")
            >>> results = model("path/to/image.jpg")
            >>> for result in results:
            >>>     result.save_txt("output.txt")

        Notes:
            - The file will contain one line per detection or classification with the following structure:
              - For detections: `class confidence x_center y_center width height`
              - For classifications: `confidence class_name`
              - For masks and keypoints, the specific formats will vary accordingly.
            - The function will create the output directory if it does not exist.
            - If save_conf is False, the confidence scores will be excluded from the output.
            - Existing contents of the file will not be overwritten; new results will be appended.
        """
        is_obb = self.obb is not None
        boxes = self.obb if is_obb else self.boxes
        masks = self.masks
        probs = self.probs
        kpts = self.keypoints
        texts = []
        if probs is not None:
            # Classify
            [texts.append(f"{probs.data[j]:.2f} {self.names[j]}") for j in probs.top5]
        elif boxes:
            # Detect/segment/pose
            for j, d in enumerate(boxes):
                if not d.is_track:
                    continue
                c, conf, id = int(d.cls), float(d.conf), int(d.id.item())
                line = (c, *(d.xyxyxyxyn.view(-1) if is_obb else d.xywhn.view(-1)))
                if masks:
                    seg = masks[j].xyn[0].copy().reshape(-1)  # reversed mask.xyn, (n,2) to (n*2)
                    line = (c, *seg)
                if kpts is not None:
                    kpt = torch.cat((kpts[j].xyn, kpts[j].conf[..., None]), 2) if kpts[j].has_visible else kpts[j].xyn
                    line += (*kpt.reshape(-1).tolist(),)
                line += (conf,) * save_conf + (() if id is None else (id,))
                texts.append(("%g " * len(line)).rstrip() % line)

        if texts:
            Path(txt_file).parent.mkdir(parents=True, exist_ok=True)  # make directory
            with open(txt_file, "a", encoding="utf-8") as f:
                f.writelines(text + "\n" for text in texts)

        return str(txt_file)

# overwrite the method on the class
Results.save_txt = my_save_txt

def on_predict_start(predictor, persist=False):
    """
    Initialize trackers for object tracking during prediction.
    Args:
        predictor (object): The predictor object to initialize trackers for.
        persist (bool, optional): Whether to persist the trackers if they already exist. Defaults to False.
    """
    assert predictor.custom_args.tracking_method in TRACKERS, \
        f"'{predictor.custom_args.tracking_method}' is not supported. Supported ones are {TRACKERS}"

    tracking_config = TRACKER_CONFIGS / (predictor.custom_args.tracking_method + '.yaml')
    trackers = []
    for i in range(predictor.dataset.bs):
        tracker = create_tracker(
            predictor.custom_args.tracking_method,
            tracking_config,
            predictor.custom_args.reid_model,
            predictor.device,
            predictor.custom_args.half,
            predictor.custom_args.per_class,
        )
        # motion only models do not have
        if hasattr(tracker, "model"):
            tracker.model.warmup()
        trackers.append(tracker)

    predictor.trackers = trackers

# callback to plot trajectories on each frame
def plot_trajectories(predictor):
    # predictor.results is a list of Results, one per frame in the batch
    for i, result in enumerate(predictor.results):
        tracker = predictor.trackers[i]
        result.orig_img = tracker.plot_results(result.orig_img, predictor.custom_args.show_trajectories)
        cv2.waitKey(1)


@torch.no_grad()
def main(args):
    if args.imgsz is None:
        args.imgsz = default_imgsz(args.yolo_model)
    yolo = YOLO(
        args.yolo_model if is_ultralytics_model(args.yolo_model) else "yolov8n.pt",
    )

    results = yolo.track(
        source=args.source,
        conf=args.conf,
        iou=args.iou,
        agnostic_nms=args.agnostic_nms,
        show=True,
        stream=True,
        device=args.device,
        show_conf=args.show_conf,
        save_txt=args.save_txt,
        show_labels=args.show_labels,
        save=args.save,
        verbose=args.verbose,
        exist_ok=args.exist_ok,
        project=args.project,
        name=args.name,
        classes=args.classes,
        imgsz=args.imgsz,
        vid_stride=args.vid_stride,
        line_width=args.line_width,
        save_crop=args.save_crop,
    )

    yolo.add_callback("on_predict_start", partial(on_predict_start, persist=True))
    yolo.add_callback("on_predict_postprocess_end", plot_trajectories)

    if not is_ultralytics_model(args.yolo_model):
        # replace yolov8 model
        m = get_yolo_inferer(args.yolo_model)
        yolo_model = m(
            model=args.yolo_model,
            device=yolo.predictor.device,
            args=yolo.predictor.args,
        )
        yolo.predictor.model = yolo_model

        # If current model is YOLOX, change the preprocess and postprocess
        if not is_ultralytics_model(args.yolo_model):
            # add callback to save image paths for further processing
            yolo.add_callback(
                "on_predict_batch_start", lambda p: yolo_model.update_im_paths(p)
            )
            yolo.predictor.preprocess = lambda imgs: yolo_model.preprocess(im=imgs)
            yolo.predictor.postprocess = lambda preds, im, im0s: yolo_model.postprocess(
                preds=preds, im=im, im0s=im0s
            )

    # store custom args in predictor
    yolo.predictor.custom_args = args

    for _ in results:
        pass


if __name__ == "__main__":
    main()

[treyrachall10]

I don't understand, will the command prompt you provided me earlier save the information to the .txt files for the provided source?

[mikel-brostrom]

An easy way of enabling this is by mokey-patching ultralytics' save_txt, this will work:

# Mikel Broström 🔥 Yolo Tracking 🧾 AGPL-3.0 license

import argparse
from functools import partial
from pathlib import Path

import cv2
import torch

from boxmot import TRACKERS
from boxmot.tracker_zoo import create_tracker
from boxmot.utils import ROOT, TRACKER_CONFIGS, WEIGHTS
from boxmot.utils.checks import RequirementsChecker
from boxmot.engine.detectors import default_imgsz, get_yolo_inferer, is_ultralytics_model

checker = RequirementsChecker()
checker.check_packages(("ultralytics", ))  # install

from ultralytics import YOLO
from ultralytics.utils.plotting import Annotator  # ultralytics.yolo.utils.plotting is deprecated
from ultralytics.utils.plotting import colors
from ultralytics.utils import plotting

# Make every drawing call a no-op
plotting.Annotator.box       = lambda *args, **kwargs: None
plotting.Annotator.box_label = lambda *args, **kwargs: None
plotting.Annotator.line      = lambda *args, **kwargs: None


from pathlib import Path
from ultralytics.engine.results import Results

def save_txt(self, txt_file: Union[str, Path], save_conf: bool = False) -> str:
        """
        Save detection results to a text file.

        Args:
            txt_file (str | Path): Path to the output text file.
            save_conf (bool): Whether to include confidence scores in the output.

        Returns:
            (str): Path to the saved text file.

        Examples:
            >>> from ultralytics import YOLO
            >>> model = YOLO("yolo11n.pt")
            >>> results = model("path/to/image.jpg")
            >>> for result in results:
            >>>     result.save_txt("output.txt")

        Notes:
            - The file will contain one line per detection or classification with the following structure:
              - For detections: `class confidence x_center y_center width height`
              - For classifications: `confidence class_name`
              - For masks and keypoints, the specific formats will vary accordingly.
            - The function will create the output directory if it does not exist.
            - If save_conf is False, the confidence scores will be excluded from the output.
            - Existing contents of the file will not be overwritten; new results will be appended.
        """
        is_obb = self.obb is not None
        boxes = self.obb if is_obb else self.boxes
        masks = self.masks
        probs = self.probs
        kpts = self.keypoints
        texts = []
        if probs is not None:
            # Classify
            [texts.append(f"{probs.data[j]:.2f} {self.names[j]}") for j in probs.top5]
        elif boxes:
            # Detect/segment/pose
            for j, d in enumerate(boxes):
                if not d.is_track:
                    continue
                c, conf, id = int(d.cls), float(d.conf), int(d.id.item())
                line = (c, *(d.xyxyxyxyn.view(-1) if is_obb else d.xywhn.view(-1)))
                if masks:
                    seg = masks[j].xyn[0].copy().reshape(-1)  # reversed mask.xyn, (n,2) to (n*2)
                    line = (c, *seg)
                if kpts is not None:
                    kpt = torch.cat((kpts[j].xyn, kpts[j].conf[..., None]), 2) if kpts[j].has_visible else kpts[j].xyn
                    line += (*kpt.reshape(-1).tolist(),)
                line += (conf,) * save_conf + (() if id is None else (id,))
                texts.append(("%g " * len(line)).rstrip() % line)

        if texts:
            Path(txt_file).parent.mkdir(parents=True, exist_ok=True)  # make directory
            with open(txt_file, "a", encoding="utf-8") as f:
                f.writelines(text + "\n" for text in texts)

        return str(txt_file)

# overwrite the method on the class
Results.save_txt = my_save_txt

def on_predict_start(predictor, persist=False):
    """
    Initialize trackers for object tracking during prediction.
    Args:
        predictor (object): The predictor object to initialize trackers for.
        persist (bool, optional): Whether to persist the trackers if they already exist. Defaults to False.
    """
    assert predictor.custom_args.tracking_method in TRACKERS, \
        f"'{predictor.custom_args.tracking_method}' is not supported. Supported ones are {TRACKERS}"

    tracking_config = TRACKER_CONFIGS / (predictor.custom_args.tracking_method + '.yaml')
    trackers = []
    for i in range(predictor.dataset.bs):
        tracker = create_tracker(
            predictor.custom_args.tracking_method,
            tracking_config,
            predictor.custom_args.reid_model,
            predictor.device,
            predictor.custom_args.half,
            predictor.custom_args.per_class,
        )
        # motion only models do not have
        if hasattr(tracker, "model"):
            tracker.model.warmup()
        trackers.append(tracker)

    predictor.trackers = trackers

# callback to plot trajectories on each frame
def plot_trajectories(predictor):
    # predictor.results is a list of Results, one per frame in the batch
    for i, result in enumerate(predictor.results):
        tracker = predictor.trackers[i]
        result.orig_img = tracker.plot_results(result.orig_img, predictor.custom_args.show_trajectories)
        cv2.waitKey(1)


@torch.no_grad()
def main(args):
    if args.imgsz is None:
        args.imgsz = default_imgsz(args.yolo_model)
    yolo = YOLO(
        args.yolo_model if is_ultralytics_model(args.yolo_model) else "yolov8n.pt",
    )

    results = yolo.track(
        source=args.source,
        conf=args.conf,
        iou=args.iou,
        agnostic_nms=args.agnostic_nms,
        show=True,
        stream=True,
        device=args.device,
        show_conf=args.show_conf,
        save_txt=args.save_txt,
        show_labels=args.show_labels,
        save=args.save,
        verbose=args.verbose,
        exist_ok=args.exist_ok,
        project=args.project,
        name=args.name,
        classes=args.classes,
        imgsz=args.imgsz,
        vid_stride=args.vid_stride,
        line_width=args.line_width,
        save_crop=args.save_crop,
    )

    yolo.add_callback("on_predict_start", partial(on_predict_start, persist=True))
    yolo.add_callback("on_predict_postprocess_end", plot_trajectories)

    if not is_ultralytics_model(args.yolo_model):
        # replace yolov8 model
        m = get_yolo_inferer(args.yolo_model)
        yolo_model = m(
            model=args.yolo_model,
            device=yolo.predictor.device,
            args=yolo.predictor.args,
        )
        yolo.predictor.model = yolo_model

        # If current model is YOLOX, change the preprocess and postprocess
        if not is_ultralytics_model(args.yolo_model):
            # add callback to save image paths for further processing
            yolo.add_callback(
                "on_predict_batch_start", lambda p: yolo_model.update_im_paths(p)
            )
            yolo.predictor.preprocess = lambda imgs: yolo_model.preprocess(im=imgs)
            yolo.predictor.postprocess = lambda preds, im, im0s: yolo_model.postprocess(
                preds=preds, im=im, im0s=im0s
            )

    # store custom args in predictor
    yolo.predictor.custom_args = args

    for _ in results:
        pass


if __name__ == "__main__":
    main()

This will