SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.05683 · 2020

TubeTK: Adopting Tubes to Track Multi-Object in a One-Step Training Model

arXiv · PDF · Open in the Atlas

Code that ran

10 of 13 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; 0 honoured the contract we drafted, 0 violated it and 9 have no contract verdict; 1 failure of our instrument; 3 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/2006.05683.json · badge: /api/ran/2006.05683.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
BoPang1996/TubeTK canonical (classifier label)
classifier basis: census
10 ran · 3 no recorded run
FunctionStatusWhere it lives
boxiou Ran
ran on a synthesized input
BoPang1996/TubeTK/Metrics.py
code served (permissive licence) · get_code("5bc06abceed1075f")
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 float64, ndim 0, 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 numpy as np

ARGS = {
    'x1': 10.0,
    'y1': 20.0,
    'w1': 30.0,
    'h1': 40.0,
    'x2': 25.0,
    'y2': 35.0,
    'w2': 35.0,
    'h2': 45.0,
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
distance2bbox Ran
ran on a synthesized input
BoPang1996/TubeTK/network/utils.py
code served (permissive licence) · get_code("878384da3a1a8ee7")
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 Tensor, ndim 2, dtype family float, shape (2, 15) 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)

# Create points tensor of shape (n, 3) where n=2
# Format: [t, x, y]
points = torch.tensor([
    [10.0, 50.0, 60.0],
    [15.0, 75.0, 85.0]
], dtype=torch.float32)

# Create distance tensor of shape (n, 14)
# Format: [left, top, right, bottom, frDis, 4point, bkDis, 4point]
# That's 14 values per point
distance = torch.tensor([
    [5.0, 10.0, 15.0, 20.0, 2.0, 1.0, 2.0, 3.0, 4.0, 1.0, 2.0, 3.0, 4.0, 5.0],
    [8.0, 12.0, 18.0, 22.0, 3.0, 1.5, 2.5, 3.5, 4.5, 2.0, 3.0, 4.0, 5.0, 6.0]
], dtype=torch.float32)

# Create max_shape list [depth/time, height, width]
max_shape = [100, 200, 300]

ARGS = {
    "points": points,
    "distance": distance,
    "max_shape": max_shape
}
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 Tensor, ndim 2, dtype family float, shape (2, 15) — shape (2, 15) 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
focal_loss Ran
ran on a synthesized input
BoPang1996/TubeTK/network/focal_loss.py
code served (permissive licence) · get_code("0d548c0e307771fa")
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 Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
from torch.autograd import Variable
import torch.nn.functional as F

torch.manual_seed(0)

# Create x: sized [N, D] where N=2 (batch), D=4 (dimensions)
# This is the model output (logits)
x = torch.randn(2, 4)

# Create y: sized [N,] where N=2 (batch)
# These are class labels (integers from 0 to num_classes-1)
# Since x.shape[1] + 1 = 5, num_classes = 5, so labels should be in range [0, 4]
y = torch.tensor([0, 2], dtype=torch.long)

ARGS = {
    'x': x,
    'y': y,
}
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 Tensor, 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
get_inter_box Ran
ran on a synthesized input
BoPang1996/TubeTK/Visualization/Vis_Input.py
code served (permissive licence) · get_code("d516af542dfff21d")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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 numpy as np

torch.manual_seed(0)

# The function performs linear interpolation between two boxes
# It takes start_box and end_box (which should be tensor-like and support arithmetic operations)
# and inter_id, end_id which are numeric identifiers

# Create small tensor boxes with shape (2, 4) - batch size 2, 4 features per box
start_box = torch.tensor([[1.0, 2.0, 3.0, 4.0],
                          [5.0, 6.0, 7.0, 8.0]])

end_box = torch.tensor([[10.0, 20.0, 30.0, 40.0],
                        [50.0, 60.0, 70.0, 80.0]])

# inter_id and end_id are numeric values used for interpolation
inter_id = 5
end_id = 10

ARGS = {
    'start_box': start_box,
    'end_box': end_box,
    'inter_id': inter_id,
    'end_id': end_id
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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_seq_info_from_file_mot Ran
ran on a synthesized input
BoPang1996/TubeTK/Visualization/Vis_Res.py
code served (permissive licence) · get_code("66cc6e4ab68b6398")
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 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 os
import tempfile
import configparser

# Create a temporary directory structure with a seqinfo.ini file
tmpdir = tempfile.mkdtemp()
seqName = "test_sequence"
seqFolder = os.path.join(tmpdir, seqName)
os.makedirs(seqFolder, exist_ok=True)

# Create a seqinfo.ini file with the required fields
seqinfo_path = os.path.join(seqFolder, "seqinfo.ini")
config = configparser.ConfigParser()
config['Sequence'] = {
    'imDir': 'img1',
    'frameRate': '30.0',
    'seqLength': '100',
    'imWidth': '1920',
    'imHeight': '1080',
    'imExt': '.jpg'
}

with open(seqinfo_path, 'w') as f:
    config.write(f)

ARGS = {
    'seqName': seqName,
    'dataDir': tmpdir
}
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 tuple, ndim 1, dtype family float — 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
giou_loss Ran
ran on a synthesized input
BoPang1996/TubeTK/network/utils.py
code served (permissive licence) · get_code("cc0dce2102e162c9")
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 Tensor, 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 torch
import numpy as np

torch.manual_seed(0)

# The function tube_giou expects:
# - pred_tubes: tensor of shape (N, 15) where each tube has:
#   [mid_t, mid_x1, mid_y1, mid_x2, mid_y2, fr_t, fr_x1, fr_y1, fr_x2, fr_y2, bk_t, bk_x1, bk_y1, bk_x2, bk_y2]
# - target_tubes: same shape as pred_tubes

# Create small batch of tubes (N=2)
# Each tube has 15 values: mid_t + 4 bbox coords + fr_t + 4 bbox coords + bk_t + 4 bbox coords

pred_tubes = torch.tensor([
    [5.0, 10.0, 20.0, 30.0, 40.0,  # mid frame: t=5, bbox=(10,20,30,40)
     7.0, 12.0, 22.0, 32.0, 42.0,  # front frame: t=7, bbox=(12,22,32,42)
     3.0, 8.0, 18.0, 28.0, 38.0],  # back frame: t=3, bbox=(8,18,28,38)
    
    [4.0, 15.0, 25.0, 35.0, 45.0,  # mid frame: t=4, bbox=(15,25,35,45)
     6.0, 17.0, 27.0, 37.0, 47.0,  # front frame: t=6, bbox=(17,27,37,47)
     2.0, 13.0, 23.0, 33.0, 43.0]  # back frame: t=2, bbox=(13,23,33,43)
], dtype=torch.float32)

target_tubes = torch.tensor([
    [5.0, 11.0, 21.0, 31.0, 41.0,  # mid frame: t=5, bbox=(11,21,31,41)
     7.0, 13.0, 23.0, 33.0, 43.0,  # front frame: t=7, bbox=(13,23,33,43)
     3.0, 9.0, 19.0, 29.0, 39.0],  # back frame: t=3, bbox=(9,19,29,39)
    
    [4.0, 16.0, 26.0, 36.0, 46.0,  # mid frame: t=4, bbox=(16,26,36,46)
     6.0, 18.0, 28.0, 38.0, 48.0,  # front frame: t=6, bbox=(18,28,38,48)
     2.0, 14.0, 24.0, 34.0, 44.0]  # back frame: t=2, bbox=(14,24,34,44)
], dtype=torch.float32)

ARGS = {
    'pred_tubes': pred_tubes,
    'target_tubes': target_tubes
}
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 Tensor, 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_loss Ran
ran on a synthesized input
BoPang1996/TubeTK/network/utils.py
code served (permissive licence) · get_code("1bcb91317c367710")
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 Tensor, 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 torch
import numpy as np

torch.manual_seed(0)

# Create valid tube tensors
# Each tube has 15 values: [mid_t, mid_bbox(4), fr_t, fr_bbox(4), bk_t, bk_bbox(4)]
# mid_t, fr_t, bk_t are temporal indices
# Each bbox is [x1, y1, x2, y2]

batch_size = 2

# Prediction tubes: 2 tubes with valid bounding boxes
pred_tubes = torch.tensor([
    [2.0, 10.0, 20.0, 50.0, 60.0,  3.0, 12.0, 22.0, 52.0, 62.0,  1.0, 8.0, 18.0, 48.0, 58.0],
    [3.0, 15.0, 25.0, 55.0, 65.0,  4.0, 17.0, 27.0, 57.0, 67.0,  2.0, 13.0, 23.0, 53.0, 63.0],
], dtype=torch.float32)

# Target tubes: 2 tubes with valid bounding boxes
target_tubes = torch.tensor([
    [2.0, 11.0, 21.0, 51.0, 61.0,  3.0, 13.0, 23.0, 53.0, 63.0,  1.0, 9.0, 19.0, 49.0, 59.0],
    [3.0, 16.0, 26.0, 56.0, 66.0,  4.0, 18.0, 28.0, 58.0, 68.0,  2.0, 14.0, 24.0, 54.0, 64.0],
], dtype=torch.float32)

ARGS = {
    "pred_tubes": pred_tubes,
    "target_tubes": target_tubes,
}
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 Tensor, 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
linear_sum_assignment_with_inf Ran
ran on a synthesized input
BoPang1996/TubeTK/Metrics.py
code served (permissive licence) · get_code("f64ce931dc5fac87")
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 2, 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 numpy as np
from scipy.optimize import linear_sum_assignment

# Create a cost matrix for the function
# The function expects a 2D array-like object
# Let's create a small 4x3 cost matrix with some finite values
cost_matrix = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0],
    [7.0, 8.0, 9.0],
    [10.0, 11.0, 12.0]
], dtype=float)

ARGS = {
    'cost_matrix': cost_matrix
}
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, ndim 2, 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
one_hot_embedding Ran
ran; our contract draft was wrong, not the code
BoPang1996/TubeTK/network/focal_loss.py
code served (permissive licence) · get_code("49cfe0b5c32b3793")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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)

# Create a LongTensor of class labels with batch size 2
# Labels should be valid indices into num_classes (0-3)
labels = torch.tensor([0, 2], dtype=torch.long)

# Number of classes
num_classes = 4

ARGS = {
    "labels": labels,
    "num_classes": num_classes,
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
parseSequences2 Ran
ran on a synthesized input
BoPang1996/TubeTK/Metrics.py
code served (permissive licence) · get_code("333a09225b53921f")
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 os
import tempfile

# Create a temporary directory and seqmap file
temp_dir = tempfile.mkdtemp()
seqmap_file = os.path.join(temp_dir, "test.seqmap")

# Write some test sequences to the file
with open(seqmap_file, 'w') as f:
    f.write("sequence1\n")
    f.write("sequence2\n")
    f.write("sequence3\n")

ARGS = {
    'seqmapFile': seqmap_file
}
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 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
multi_apply Not yet run
unverified (harvested, no recorded run)
BoPang1996/TubeTK/network/track_head.py
code served (permissive licence) · get_code("f7898a5762a4d3e4")
How it ran

no invocation recorded.

tubes2bbox Not yet run
unverified (harvested, no recorded run)
BoPang1996/TubeTK/Visualization/Vis_Input.py
code served (permissive licence) · get_code("2c479e738ef20c41")
How it ran

no invocation recorded.

tubes2bbox_out Not yet run
unverified (harvested, no recorded run)
BoPang1996/TubeTK/Visualization/Vis_Input.py
code served (permissive licence) · get_code("17ad0accc3f687bf")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 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
multi_apply 10 other papers
2 dated earlier, 1 the same month, 7 later
1911.09070 (2019-11)
EfficientDet: Scalable and Efficient Object Detection
2003.10152 (2020-03)
SOLOv2: Dynamic and Fast Instance Segmentation
2006.11275 (2020-06)
Center-based 3D Object Detection and Tracking
and 7 more not named here
one_hot_embedding 1 other paper
1 later
2310.08255 (2023-10)
Leveraging Vision-Language Models for Improving Domain Generalization in Image Classification

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

Multi-object tracking is a fundamental vision problem that has been studied for a long time. As deep learning brings excellent performances to object detection algorithms, Tracking by Detection (TBD) has become the mainstream tracking framework. Despite the success of TBD, this two-step method is too complicated to train in an end-to-end manner and induces many challenges as well, such as insufficient exploration of video spatial-temporal information, vulnerability when facing object occlusion, and excessive reliance on detection results. To address these challenges, we propose a concise end-to-end model TubeTK which only needs one step training by introducing the ``bounding-tube" to indicate temporal-spatial locations of objects in a short video clip. TubeTK provides a novel direction of multi-object tracking, and we demonstrate its potential to solve the above challenges without bells and whistles. We analyze the performance of TubeTK on several MOT benchmarks and provide empirical evidence to show that TubeTK has the ability to overcome occlusions to some extent without any ancillary technologies like Re-ID. Compared with other methods that adopt private detection results, our one-stage end-to-end model achieves state-of-the-art performances even if it adopts no ready-made detection results. We hope that the proposed TubeTK model can serve as a simple but strong alternative for video-based MOT task. The code and models are available at https://github.com/BoPang1996/TubeTK.

For agents

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

get_harvested_code_for_paper("2006.05683")
get_code_for_paper("2006.05683")
have("2006.05683")

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

curl https://syntology.ai/api/ran/2006.05683.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.05683.svg)](https://syntology.ai/paper/2006.05683)

Connect an agent — have() is free.