SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.13514 · 2022

StrongSORT: Make DeepSORT Great Again

arXiv · PDF · Open in the Atlas

Code that ran

16 of 20 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 11 have no contract verdict; 3 failures of our instrument; 4 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2202.13514.json · badge: /api/ran/2202.13514.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
dyhBUPT/StrongSORT canonical (classifier label)
classifier basis: classifier_2025
1 ran · 1 no recorded run
goksenin-uav/strongsort-pip community (classifier label)
classifier basis not recorded on the link
14 ran · 0 no recorded run
kadirnar/deepsort-pip community (classifier label)
classifier basis not recorded on the link
1 ran · 3 no recorded run
FunctionStatusWhere it lives
LinearInterpolation Ran
ran, honoured the contract we drafted
dyhBUPT/StrongSORT/GSI.py
pointer only (licence: GPL-3.0) · get_code("291f44f2e2225f0d")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, dyhBUPT/StrongSORT, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

assert_in Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/asserts.py
code served (permissive licence) · get_code("287f12a551c1c76a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function checks if 'file' is in 'files_to_check'
# Both parameters appear to be compared for membership, so they can be strings or other comparable objects

file = "model.pth"
files_to_check = ["model.pth", "config.json", "weights.bin"]

ARGS = {
    "file": file,
    "files_to_check": files_to_check
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
assert_in_env Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/asserts.py
code served (permissive licence) · get_code("3d29e11e107d75b2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
from os import environ

# Set up some environment variables for the function to check
os.environ['TEST_VAR_1'] = 'value1'
os.environ['TEST_VAR_2'] = 'value2'
os.environ['TEST_VAR_3'] = 'value3'

# Define ARGS with a check_list parameter
ARGS = {
    'check_list': ['TEST_VAR_1', 'TEST_VAR_2', 'TEST_VAR_3']
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_color_for_labels Ran
ran, honoured the contract we drafted
goksenin-uav/strongsort-pip/strongsort/draw.py
pointer only (licence: MIT); the stored copy get_code serves, from mikel-brostrom/Yolov3_DeepSort_Pytorch, is pointer only (licence: NONE) · get_code("bcfa4cf98c4b792a")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from mikel-brostrom/Yolov3_DeepSort_Pytorch, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

draw_boxes Ran
ran; our contract draft was wrong, not the code
goksenin-uav/strongsort-pip/strongsort/draw.py
code served (permissive licence) · get_code("bb6bb0009b53e1ba")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family int, shape (480, 640, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import cv2
import numpy as np

# Create a small test image
img = np.zeros((480, 640, 3), dtype=np.uint8)

# Create bounding boxes in format [x1, y1, x2, y2]
# Using 4 boxes with distinct coordinates
bbox = np.array([
    [10, 20, 100, 150],
    [200, 100, 350, 250],
    [400, 50, 550, 180],
    [50, 300, 200, 400]
], dtype=np.float32)

# Create identity labels for each box
identities = np.array([1, 2, 5, 10], dtype=np.int32)

# Define offset
offset = (0, 0)

ARGS = {
    "img": img,
    "bbox": bbox,
    "identities": identities,
    "offset": offset
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (480, 640, 3) — shape (480, 640, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_logger Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/log.py
code served (permissive licence) · get_code("1fcd8693108efe91")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Logger. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import logging

def get_logger(name='root'):
    formatter = logging.Formatter(fmt='%(asctime)s [%(levelname)s]: %(message)s', datefmt='%Y-%m-%d %H:%M:%S')
    handler = logging.StreamHandler()
    handler.setFormatter(formatter)
    logger = logging.getLogger(name)
    logger.setLevel(logging.INFO)
    logger.addHandler(handler)
    return logger

ARGS = {
    'name': 'test_logger'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Logger — a Logger output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Logger output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_model_name Ran
ran; our contract draft was wrong, not the code
goksenin-uav/strongsort-pip/strongsort/reid_model_factory.py
pointer only (licence: MIT); the stored copy get_code serves, from jacobtyo/mudd, is pointer only (licence: NONE) · get_code("c5cdc64a6b2b4eee")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from jacobtyo/mudd, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_model_url Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/reid_model_factory.py
code served (permissive licence) · get_code("2745b5d88394ba76")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

# The function takes a 'model' parameter and converts it to string,
# then checks if it's in the __trained_urls dictionary.
# We need to provide a model that when converted to string and processed
# will match one of the keys in __trained_urls.

# Simple test with a string that matches a key in the dictionary
model_input = 'resnet50_market1501.pt'

ARGS = {
    'model': model_input
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
iou Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/iou_matching.py
code served (permissive licence) · get_code("cd5390257debcfd6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (2,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a bounding box in format (top left x, top left y, width, height)
bbox = np.array([10.0, 20.0, 30.0, 40.0])

# Create candidate bounding boxes (multiple boxes, same format)
candidates = np.array([
    [12.0, 22.0, 25.0, 35.0],
    [5.0, 15.0, 50.0, 50.0],
])

ARGS = {
    "bbox": bbox,
    "candidates": candidates,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (2,) — shape (2,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
iou_cost Ran
ran on a synthesized input
kadirnar/deepsort-pip/deepsort/iou_matching.py
code served (permissive licence) · get_code("fa2c5a05698eba02")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (2, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

class MockTrack:
    def __init__(self, tlwh, time_since_update=0):
        self.tlwh_val = tlwh
        self.time_since_update = time_since_update
    
    def to_tlwh(self):
        return self.tlwh_val

class MockDetection:
    def __init__(self, tlwh):
        self.tlwh = tlwh

class MockLinearAssignment:
    INFTY_COST = 1e5

linear_assignment = MockLinearAssignment()

tracks = [
    MockTrack(np.array([10.0, 20.0, 50.0, 100.0]), time_since_update=0),
    MockTrack(np.array([15.0, 25.0, 60.0, 110.0]), time_since_update=0),
]

detections = [
    MockDetection(np.array([12.0, 22.0, 55.0, 105.0])),
    MockDetection(np.array([100.0, 200.0, 40.0, 80.0])),
    MockDetection(np.array([20.0, 30.0, 50.0, 100.0])),
]

ARGS = {
    "tracks": tracks,
    "detections": detections,
    "track_indices": None,
    "detection_indices": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 3) — shape (2, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
is_model_in_model_types Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/reid_model_factory.py
code served (permissive licence) · get_code("89a76bc51c8b9c46")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

# The function is_model_in_model_types takes a model parameter
# and converts it to string, then extracts the filename (last part after '/') 
# and the base name (before first '.'), then checks if it's in __model_types

# Valid test cases that should return True
valid_model_name_1 = "resnet50"
valid_model_name_2 = "resnet50.pth"
valid_model_name_3 = "/path/to/resnet50.pth"
valid_model_name_4 = "/some/path/mobilenetv2_x1_0.pth"

# We can pass any object that when converted to string produces a valid model name
# Let's create a simple object that stringifies to a valid model path
class ModelPath:
    def __init__(self, path):
        self.path = path
    
    def __str__(self):
        return self.path

# Create test model path
model = ModelPath("resnet50.pth")

# Alternatively, we can just pass a string directly
model = "resnet50.pth"

ARGS = {
    "model": model
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
matching_cascade Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/linear_assignment.py
code served (permissive licence) · get_code("402e8c9566232aa2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

class Track:
    """Minimal Track class for matching_cascade"""
    def __init__(self, track_id, age):
        self.track_id = track_id
        self.age = age

class Detection:
    """Minimal Detection class for matching_cascade"""
    def __init__(self, detection_id, feature):
        self.detection_id = detection_id
        self.feature = feature

def distance_metric_fn(tracks, detections, track_indices, detection_indices):
    """
    Simple distance metric that returns cost matrix.
    Returns NxM matrix where N=len(track_indices), M=len(detection_indices)
    """
    n_tracks = len(track_indices)
    n_detections = len(detection_indices)
    cost_matrix = np.random.rand(n_tracks, n_detections) * 5.0
    return cost_matrix

tracks = [
    Track(track_id=0, age=1),
    Track(track_id=1, age=3),
    Track(track_id=2, age=5),
]

detections = [
    Detection(detection_id=0, feature=np.array([1.0, 2.0, 3.0, 4.0])),
    Detection(detection_id=1, feature=np.array([2.0, 3.0, 4.0, 5.0])),
    Detection(detection_id=2, feature=np.array([3.0, 4.0, 5.0, 6.0])),
]

ARGS = {
    "distance_metric": distance_metric_fn,
    "max_distance": 2.5,
    "cascade_depth": 5,
    "tracks": tracks,
    "detections": detections,
    "track_indices": None,
    "detection_indices": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
min_cost_matching Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/linear_assignment.py
code served (permissive licence) · get_code("6278dd3b36b3eadc")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Define minimal stand-in classes for Track and Detection
class Track:
    def __init__(self, track_id, state):
        self.track_id = track_id
        self.state = state

class Detection:
    def __init__(self, detection_id, feature):
        self.detection_id = detection_id
        self.feature = feature

# Define a distance metric function that computes cost matrix
def distance_metric(tracks, detections, track_indices, detection_indices):
    """
    Compute NxM cost matrix where N = len(track_indices), M = len(detection_indices).
    Returns a matrix with distinct dimensions.
    """
    n_tracks = len(track_indices)
    n_detections = len(detection_indices)
    # Create a simple cost matrix with values between 0 and 10
    cost_matrix = np.random.rand(n_tracks, n_detections) * 8.0
    return cost_matrix

# Create sample tracks and detections
tracks = [
    Track(0, np.array([1.0, 2.0, 3.0, 4.0])),
    Track(1, np.array([5.0, 6.0, 7.0, 8.0])),
    Track(2, np.array([9.0, 10.0, 11.0, 12.0])),
]

detections = [
    Detection(0, np.array([1.1, 2.1, 3.1, 4.1])),
    Detection(1, np.array([5.1, 6.1, 7.1, 8.1])),
]

# Define arguments
ARGS = {
    "distance_metric": distance_metric,
    "max_distance": 5.0,
    "tracks": tracks,
    "detections": detections,
    "track_indices": None,
    "detection_indices": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
read_mot_results Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/io.py
code served (permissive licence) · get_code("9eb3b334fd02b5d3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
import tempfile

# Create a temporary MOT-format results file
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "MOT17-02-gt.txt")

# Write sample MOT format data
# Format: FrameID, TargetID, X, Y, W, H, Mark, Label, Visibility
mot_data = """1,1,10.5,20.3,50.0,100.0,1,1,0.95
1,2,200.0,150.0,45.0,95.0,1,1,0.88
2,1,12.0,21.5,50.5,100.5,1,1,0.96
2,3,300.0,200.0,40.0,90.0,1,2,0.75
3,1,15.0,25.0,51.0,101.0,1,1,0.98
"""

with open(temp_file, 'w') as f:
    f.write(mot_data)

ARGS = {
    "filename": temp_file,
    "is_gt": True,
    "is_ignore": False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
read_results Ran
ran; our contract draft was wrong, not the code
goksenin-uav/strongsort-pip/strongsort/io.py
code served (permissive licence) · get_code("1958a4fce355d7a6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
import tempfile

# Create a temporary MOT format file for testing
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_filename = temp_file.name

# Write sample MOT format data
# Format: <frame>, <id>, <x>, <y>, <w>, <h>, <conf>, <class_id>, <visibility>
# For ground truth (is_gt=True), we need MOT16/MOT17 format with proper labels
sample_data = """1,1,10.5,20.3,50.2,100.1,1,1,0.9
2,1,11.0,21.0,50.0,100.0,0.95,1,0.95
3,2,30.0,40.0,60.0,120.0,0.8,1,0.8
"""

temp_file.write(sample_data)
temp_file.close()

# Create another file for ignore testing
temp_ignore_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_ignore_filename = temp_ignore_file.name

ignore_data = """1,1,10.5,20.3,50.2,100.1,1,2,0.5
2,1,11.0,21.0,50.0,100.0,0.95,7,0.3
"""

temp_ignore_file.write(ignore_data)
temp_ignore_file.close()

# Define arguments for read_results function
ARGS = {
    'filename': temp_filename,
    'data_type': 'mot',
    'is_gt': False,
    'is_ignore': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
unzip_objs Ran
ran on a synthesized input
goksenin-uav/strongsort-pip/strongsort/io.py
code served (permissive licence) · get_code("261a9c5fae0c1b9f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# The function unzip_objs expects a list/iterable of tuples
# Each tuple should contain (tlwh, id, score)
# where tlwh is a 4-element array-like (top, left, width, height)
# id is an identifier
# score is a confidence score

# Create sample objects as tuples
obj1 = ([10, 20, 100, 150], 1, 0.95)
obj2 = ([50, 60, 200, 250], 2, 0.87)
obj3 = ([15, 25, 110, 160], 3, 0.92)

objs = [obj1, obj2, obj3]

ARGS = {
    "objs": objs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
GaussianSmooth Not yet run
unverified (harvested, no recorded run)
dyhBUPT/StrongSORT/GSI.py
pointer only (licence: GPL-3.0) · get_code("e8e20ac10a09b701")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, dyhBUPT/StrongSORT, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

gate_cost_matrix Not yet run
unverified (harvested, no recorded run)
kadirnar/deepsort-pip/deepsort/linear_assignment.py
code served (permissive licence) · get_code("fd534ce01d39359c")
How it ran

no invocation recorded.

matching_cascade Not yet run
unverified (harvested, no recorded run)
kadirnar/deepsort-pip/deepsort/linear_assignment.py
code served (permissive licence) · get_code("8a6c0862c2243eaa")
How it ran

no invocation recorded.

min_cost_matching Not yet run
unverified (harvested, no recorded run)
kadirnar/deepsort-pip/deepsort/linear_assignment.py
code served (permissive licence) · get_code("dbbe07d4ffcebac6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 20 samples lifted from this paper's linked repositories, 20 carry a function identity and 0 do not. Of the 20 distinct functions with an identity, 14 are also present in other papers whose code we hold (14 with at least one paper dated earlier, 0 the same month, 0 later) and 6 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
read_mot_results 4 other papers
4 dated earlier
1909.12605 (2019-09)
Towards Real-Time Multi-Object Tracking
2004.01888 (2020-04)
FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Tracking
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
and 1 more not named here
read_results 4 other papers
4 dated earlier
1909.12605 (2019-09)
Towards Real-Time Multi-Object Tracking
2004.01888 (2020-04)
FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Tracking
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
and 1 more not named here
unzip_objs 4 other papers
4 dated earlier
1909.12605 (2019-09)
Towards Real-Time Multi-Object Tracking
2004.01888 (2020-04)
FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Tracking
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
and 1 more not named here
compute_color_for_labels 2 other papers
2 dated earlier
1703.07402 (2017-03)
Simple Online and Realtime Tracking with a Deep Association Metric
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
draw_boxes 2 other papers
2 dated earlier
1703.07402 (2017-03)
Simple Online and Realtime Tracking with a Deep Association Metric
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
gate_cost_matrix 2 other papers
2 dated earlier
1703.07402 (2017-03)
Simple Online and Realtime Tracking with a Deep Association Metric
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
iou 2 other papers
2 dated earlier
1703.07402 (2017-03)
Simple Online and Realtime Tracking with a Deep Association Metric
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
iou_cost 2 other papers
2 dated earlier
1703.07402 (2017-03)
Simple Online and Realtime Tracking with a Deep Association Metric
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
assert_in 1 other paper
1 dated earlier
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
assert_in_env 1 other paper
1 dated earlier
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
get_logger 1 other paper
1 dated earlier
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
get_model_name 1 other paper
1 dated earlier
1910.06827 (2019-10)
Learning Generalisable Omni-Scale Representations for Person Re-Identification
matching_cascade 1 other paper
1 dated earlier
1703.07402 (2017-03)
Simple Online and Realtime Tracking with a Deep Association Metric
min_cost_matching 1 other paper
1 dated earlier
2104.14294 (2021-04)
Emerging Properties in Self-Supervised Vision Transformers

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Recently, Multi-Object Tracking (MOT) has attracted rising attention, and accordingly, remarkable progresses have been achieved. However, the existing methods tend to use various basic models (e.g, detector and embedding model), and different training or inference tricks, etc. As a result, the construction of a good baseline for a fair comparison is essential. In this paper, a classic tracker, i.e., DeepSORT, is first revisited, and then is significantly improved from multiple perspectives such as object detection, feature embedding, and trajectory association. The proposed tracker, named StrongSORT, contributes a strong and fair baseline for the MOT community. Moreover, two lightweight and plug-and-play algorithms are proposed to address two inherent "missing" problems of MOT: missing association and missing detection. Specifically, unlike most methods, which associate short tracklets into complete trajectories at high computation complexity, we propose an appearance-free link model (AFLink) to perform global association without appearance information, and achieve a good balance between speed and accuracy. Furthermore, we propose a Gaussian-smoothed interpolation (GSI) based on Gaussian process regression to relieve the missing detection. AFLink and GSI can be easily plugged into various trackers with a negligible extra computational cost (1.7 ms and 7.1 ms per image, respectively, on MOT17). Finally, by fusing StrongSORT with AFLink and GSI, the final tracker (StrongSORT++) achieves state-of-the-art results on multiple public benchmarks, i.e., MOT17, MOT20, DanceTrack and KITTI. Codes are available at https://github.com/dyhBUPT/StrongSORT and https://github.com/open-mmlab/mmtracking.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2202.13514")
get_code_for_paper("2202.13514")
have("2202.13514")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2202.13514.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2202.13514.svg)](https://syntology.ai/paper/2202.13514)

Connect an agent — have() is free.