SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.11186 · CVPR · 2024

NetTrack: Tracking Highly Dynamic Objects with a Net

Shijie Lin, Jia Pan, Guangze Zheng, Haobo Zuo, Changhong Fu

arXiv · PDF · Open in the Atlas

Code that ran

9 of 14 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 6 failures of our instrument; 5 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/2403.11186.json · badge: /api/ran/2403.11186.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
George-Zhuang/NetTrack community
Papers with Code archive
9 ran · 5 no recorded run
FunctionStatusWhere it lives
BaseTrack Ran
ran on a synthesized input
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("c2efbeca34d34df4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned NoneType. 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
from collections import OrderedDict

class TrackState(object):
    New = 0
    Tracked = 1
    Lost = 2
    Removed = 3

class BaseTrack(object):
    _count = 0
    track_id = 0
    is_activated = False
    state = TrackState.New
    history = OrderedDict()
    features = []
    curr_feature = None
    score = 0
    start_frame = 0
    frame_id = 0
    time_since_update = 0
    location = (np.inf, np.inf)

    @property
    def end_frame(self):
        return self.frame_id

    @staticmethod
    def next_id():
        BaseTrack._count += 1
        return BaseTrack._count

    def activate(self, *args):
        raise NotImplementedError

    def predict(self):
        raise NotImplementedError

    def update(self, *args, **kwargs):
        raise NotImplementedError

    def mark_lost(self):
        self.state = TrackState.Lost

    def mark_removed(self):
        self.state = TrackState.Removed


INIT_ARGS = {}

ARGS = {}

METHOD = 'mark_lost'

if __name__ == '__main__':
    track = BaseTrack(**INIT_ARGS)
    print(f"Initial state: {track.state}")
    print(f"Initial is_activated: {track.is_activated}")
    print(f"Initial location: {track.location}")
    
    getattr(track, METHOD)(**ARGS)
    
    print(f"State after mark_lost: {track.state}")
    print(f"State is Lost: {track.state == TrackState.Lost}")
    
    track.mark_removed()
    print(f"State after mark_removed: {track.state}")
    print(f"State is Removed: {track.state == TrackState.Removed}")
    
    next_id_1 = BaseTrack.next_id()
    next_id_2 = BaseTrack.next_id()
    print(f"Next IDs: {next_id_1}, {next_id_2}")
    print(f"End frame: {track.end_frame}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NoneType — a NoneType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoneType 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
TrackState Ran
ran on a synthesized input · constructed an object; no result was computed
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("3d5b0c832e5a2d89")
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 an instance of its own class, TrackState: the run constructed the object and did not call it, so no result was computed. 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
import torch.nn as nn

torch.manual_seed(0)

ARGS = {}
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 TrackState — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a TrackState 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_color Ran
ran, honoured the contract we drafted
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("0d2550860de1d622")
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, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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)

ARGS = {
    "idx": 5
}
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 tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
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
joint_stracks Ran
ran; our contract draft was wrong, not the code
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("557de671825f7887")
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 list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 numpy as np

class Track:
    def __init__(self, track_id):
        self.track_id = track_id

torch.manual_seed(0)

track1 = Track(track_id=1)
track2 = Track(track_id=2)
track3 = Track(track_id=3)
track4 = Track(track_id=2)

tlista = [track1, track2]
tlistb = [track3, track4]

ARGS = {
    "tlista": tlista,
    "tlistb": tlistb
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
plot_crossmark Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("fed926ad88e2707b")
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 (100, 100, 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
import torch

torch.manual_seed(0)

# Create a small image (BGR format for OpenCV)
image = np.zeros((100, 100, 3), dtype=np.uint8)

# Define coordinates as a tuple of (x, y)
coordinates = (50, 50)

# Define color as BGR tuple
color = (0, 255, 0)

# Optional parameters with defaults
sz = 8
thickness = 3

ARGS = {
    "image": image,
    "coordinates": coordinates,
    "color": color,
    "sz": sz,
    "thickness": thickness
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy, torch — 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 (100, 100, 3) — shape (100, 100, 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
plot_points Ran
ran; our contract draft was wrong, not the code
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("350b6bb21cc6959c")
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
from colorsys import hsv_to_rgb

# Create a valid image (BGR format for OpenCV)
image = np.ones((480, 640, 3), dtype=np.uint8) * 255

# Create point stacks - list of lists containing (x, y) coordinates
point_stacks = [
    [(100, 150), (200, 250)],
    [(300, 100), (400, 200), (500, 300)]
]

# Create tlwhs - list of bounding boxes in format (x, y, width, height)
tlwhs = [
    (50, 100, 200, 150),
    (250, 50, 300, 200)
]

ARGS = {
    "image": image,
    "point_stacks": point_stacks,
    "tlwhs": tlwhs,
    "point_thickness": 10,
    "crossmark_thickness": 7
}
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
plot_tracking Ran
ran; our contract draft was wrong, not the code
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("3b1ef7ad76cdfce5")
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 simple test image
image = np.random.randint(0, 255, (480, 640, 3), dtype=np.uint8)

# Create bounding boxes in tlwh format (top-left-x, top-left-y, width, height)
tlwhs = [
    [50, 50, 100, 150],
    [200, 100, 80, 120],
]

# Create object IDs
obj_ids = [1, 2]

# Create scores (optional)
scores = [0.95, 0.87]

# Create frame_id and fps
frame_id = 0
fps = 30.0

# Create ids2 (optional, for secondary IDs)
ids2 = [10, 20]

# Line thickness
line_thickness = 10

ARGS = {
    "image": image,
    "tlwhs": tlwhs,
    "obj_ids": obj_ids,
    "scores": scores,
    "frame_id": frame_id,
    "fps": fps,
    "ids2": ids2,
    "line_thickness": line_thickness,
}
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
save_result Ran
ran; our contract draft was wrong, not the code
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("7aafbf95486fb2c0")
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 numpy as np
import tempfile
import os

class MockTrack:
    def __init__(self, track_id, tlwh, score, curr_points):
        self.track_id = track_id
        self.tlwh = tlwh
        self.score = score
        self.curr_points = curr_points

# Create a temporary image file for testing
temp_dir = tempfile.mkdtemp()
img_path = os.path.join(temp_dir, "test_image.jpg")

# Create a simple test image (480x640 BGR format)
test_image = np.random.randint(0, 255, (480, 640, 3), dtype=np.uint8)
import cv2
cv2.imwrite(img_path, test_image)

# Create mock stracks
stracks = [
    MockTrack(
        track_id=1,
        tlwh=np.array([50.0, 60.0, 100.0, 150.0]),
        score=0.95,
        curr_points=[[100.0, 120.0], [110.0, 130.0]]
    ),
    MockTrack(
        track_id=2,
        tlwh=np.array([200.0, 150.0, 80.0, 120.0]),
        score=0.87,
        curr_points=[[240.0, 180.0], [250.0, 200.0]]
    ),
]

# Create output directory
output_path = os.path.join(temp_dir, "output")
os.makedirs(output_path, exist_ok=True)

# Create sequence subdirectory
seq_name = "test_seq"
os.makedirs(os.path.join(output_path, seq_name), exist_ok=True)

ARGS = {
    "img_path": img_path,
    "stracks": stracks,
    "frame_id": 0,
    "seq_name": seq_name,
    "output_path": output_path,
    "vis_box": True,
    "vis_points": True,
    "extra_points": None,
    "min_area": 10,
    "max_area": 1000000000000,
}
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 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
sub_stracks Ran
ran; our contract draft was wrong, not the code
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("322ac55b144012cf")
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 list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 numpy as np

class Track:
    def __init__(self, track_id):
        self.track_id = track_id

# Create two lists of tracks
tlista = [
    Track(track_id=1),
    Track(track_id=2),
    Track(track_id=3),
]

tlistb = [
    Track(track_id=2),
    Track(track_id=4),
]

ARGS = {
    'tlista': tlista,
    'tlistb': tlistb,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
BaseTracker Not yet run
unverified (harvested, no recorded run)
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("b6efe16f92216d0c")
How it ran

no invocation recorded.

KalmanFilter Not yet run
unverified (harvested, no recorded run)
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("684663bc0c1cf075")
How it ran

no invocation recorded.

NetTracker Not yet run
unverified (harvested, no recorded run)
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("fe14e5bfcf385b70")
How it ran

no invocation recorded.

STrack Not yet run
unverified (harvested, no recorded run)
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("721b4224302359bf")
How it ran

no invocation recorded.

remove_duplicate_stracks Not yet run
unverified (harvested, no recorded run)
George-Zhuang/NetTrack/tracker/nettrack.py
code served (permissive licence) · get_code("1d1503793d1e312b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 2 later) and 11 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
joint_stracks
brick of this paper's assembly: helper
7 other papers
4 dated earlier, 2 later, 1 same year, order not determinable
1909.12605 (2019-09)
Towards Real-Time Multi-Object Tracking
2010.12138 (2020-10)
Rethinking the competition between detection and ReID in Multi-Object Tracking
2206.14651 (2022-06)
BoT-SORT: Robust Associations Multi-Pedestrian Tracking
and 4 more not named here
sub_stracks
brick of this paper's assembly: helper
7 other papers
4 dated earlier, 2 later, 1 same year, order not determinable
1909.12605 (2019-09)
Towards Real-Time Multi-Object Tracking
2010.12138 (2020-10)
Rethinking the competition between detection and ReID in Multi-Object Tracking
2206.14651 (2022-06)
BoT-SORT: Robust Associations Multi-Pedestrian Tracking
and 4 more not named here
get_color
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2004.01888 (2020-04)
FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Tracking

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

Open-world prompt: "Osprey (Unknown) + fish (Known)" Deformation! Deformation! Referring prompt : "A large bird with brown head, wings, and white tail is predating" Fast motion! Deformation! b a Track with a Box (traditional) Track with a Net (ours) Internal relationships Failure Robust Coarse-grained Fine-grained Figure 1. a The visualization of the proposed NetTrack is similar to a Net. Object dynamicity distorts the internal relationships of the object, presenting challenges for traditional coarse-grained tracking methods that rely solely on bounding boxes. While NetTrack introduces fine-grained Nets that are robust to dynamicity. b Qualitative results of NetTrack tracking highly dynamic objects under openworld tracking and referring expression comprehension settings. Dynamicity like deformation and fast motion results in drastic changes in the coarse-grained representation, while the fine-grained Nets can contract robustly. The dashed boxes represent the object position from the previous time step. c We propose a challenging benchmark named BFT, dedicated to evaluating highly dynamic object tracking with abundant scenarios shown in the external circular and diverse species shown in the central word cloud.

For agents

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

get_harvested_code_for_paper("2403.11186")
get_code_for_paper("2403.11186")
have("2403.11186")

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

curl https://syntology.ai/api/ran/2403.11186.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.11186.svg)](https://syntology.ai/paper/2403.11186)

Connect an agent — have() is free.