SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2004.01177 · 2020

Tracking Objects as Points

arXiv · PDF · Open in the Atlas

Code that ran

10 of 16 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 5 have no contract verdict; 5 failures of our instrument; 6 no recorded run), measured 2026-11-21; 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/2004.01177.json · badge: /api/ran/2004.01177.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
xingyizhou/CenterTrack canonical (classifier label)
classifier basis: census
2 ran · 3 no recorded run
hampen2929/CenterTrack community (classifier label)
classifier basis not recorded on the link
5 ran · 0 no recorded run
qa276390/searchtrack community (classifier label)
classifier basis: plan_row
3 ran · 3 no recorded run
FunctionStatusWhere it lives
compute_bin_loss Ran
ran; our contract draft was wrong, not the code
qa276390/searchtrack/src/lib/model/losses.py
code served (permissive licence) · get_code("65d7a40d70c53ed9")
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. 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
import torch.nn.functional as F

torch.manual_seed(0)

# Analyze the function:
# - output: passed to F.cross_entropy, so should be (batch_size, num_classes) logits
# - target: passed to F.cross_entropy, so should be (batch_size,) with class indices
# - mask: has .expand_as(output) called on it, then .float() is called
#   mask should be broadcastable to output shape

batch_size = 2
num_classes = 4

# output: logits tensor (batch_size, num_classes)
output = torch.randn(batch_size, num_classes)

# target: class indices (batch_size,) with values in [0, num_classes)
target = torch.tensor([0, 2], dtype=torch.long)

# mask: should be broadcastable to output shape
# Can be a single value or smaller tensor that expands
mask = torch.ones(batch_size, 1)

ARGS = {
    "output": output,
    "target": target,
    "mask": mask,
}
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
compute_res_loss Ran
ran; our contract draft was wrong, not the code
qa276390/searchtrack/src/lib/model/losses.py
code served (permissive licence) · get_code("7df13fb9e54d592b")
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. 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.functional as F

torch.manual_seed(0)

# Create small tensors for output and target
# batch_size=2, feature_dim=4
output = torch.randn(2, 4)
target = torch.randn(2, 4)

ARGS = {
    'output': output,
    'target': target,
}
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
compute_rot_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
xingyizhou/CenterTrack/src/lib/model/losses.py
code served (permissive licence) · get_code("8accefa95d8419ec")
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. 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.functional as F

torch.manual_seed(0)

# Create output tensor: shape (batch_size, 8) since output.view(-1, 8) is called
output = torch.randn(2, 8)

# Create target_bin tensor: shape (batch_size, 2) since target_bin.view(-1, 2) is called
# Values should be 0 or 1 for cross_entropy classification
target_bin = torch.tensor([[1, 0], [0, 1]], dtype=torch.long)

# Create target_res tensor: shape (batch_size, 2) since target_res.view(-1, 2) is called
# These are angle values that will have sin/cos applied
target_res = torch.randn(2, 2)

# Create mask tensor: shape (batch_size, 1) since mask.view(-1, 1) is called
# Mask should be binary or float values
mask = torch.tensor([[1], [1]], dtype=torch.float32)

ARGS = {
    'output': output,
    'target_bin': target_bin,
    'target_res': target_res,
    'mask': mask
}
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
gen_nan_mat Ran
ran on a synthesized input
hampen2929/CenterTrack/notebook/CenterTrack.py
code served (permissive licence) · get_code("8deded4e22a90040")
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, 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 numpy as np

# Test case 1: dim only (data_num is None)
dim_1 = 5
data_num_1 = None

# Test case 2: dim with data_num specified
dim_2 = 3
data_num_2 = 7

# Use the second test case for ARGS
ARGS = {
    "dim": 4,
    "data_num": 2
}
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, 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_images Ran
ran on a synthesized input
hampen2929/CenterTrack/notebook/util.py
code served (permissive licence) · get_code("e5ab43a0ed3a9d85")
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 torch

ARGS = {
    'label_file': {
        'imagePath': 'test_image.jpg',
        'imageHeight': 480,
        'imageWidth': 640
    },
    'image_id': 12345,
    'file_name': None
}
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 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
get_json_file_paths Ran
ran on a synthesized input
hampen2929/CenterTrack/notebook/util.py
code served (permissive licence) · get_code("2e11ec34b5533614")
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
import json
from pathlib import Path

# Create a temporary directory with some JSON files
temp_dir = tempfile.mkdtemp()

# Create a few test JSON files
test_files = ['file1.json', 'file2.json', 'file3.json']
for filename in test_files:
    filepath = os.path.join(temp_dir, filename)
    with open(filepath, 'w') as f:
        json.dump({'test': 'data'}, f)

ARGS = {
    'data_dir': temp_dir
}
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
load_json Ran
ran on a synthesized input
hampen2929/CenterTrack/notebook/util.py
code served (permissive licence) · get_code("f0654f3740c34713")
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 json
import tempfile
import os

# Create a temporary JSON file with sample data
temp_dir = tempfile.mkdtemp()
json_file_path = os.path.join(temp_dir, "test_data.json")

sample_data = {
    "key1": "value1",
    "key2": [1, 2, 3],
    "key3": {"nested": "object"}
}

with open(json_file_path, 'w') as f:
    json.dump(sample_data, f)

ARGS = {
    "data_path": json_file_path
}
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
load_model Ran
ran; our contract draft was wrong, not the code
xingyizhou/CenterTrack/src/lib/model/model.py
code served (permissive licence) · get_code("d7425309b038c87c")
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) 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 torch
import torch.nn as nn
import tempfile
import os

torch.manual_seed(0)

# Create a minimal model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv = nn.Conv2d(3, 16, 3)
        self.hm = nn.Conv2d(16, 80, 1)
    
    def forward(self, x):
        return self.hm(self.conv(x))

# Create a minimal optimizer config
class OptConfig:
    def __init__(self):
        self.reset_hm = False
        self.reuse_hm = False
        self.resume = False
        self.lr = 0.001
        self.lr_step = [10, 20]

# Create a checkpoint file
model = SimpleModel()
optimizer = torch.optim.Adam(model.parameters(), lr=0.001)

checkpoint = {
    'epoch': 5,
    'state_dict': model.state_dict(),
    'optimizer': optimizer.state_dict()
}

# Write checkpoint to temporary file
temp_dir = tempfile.gettempdir()
model_path = os.path.join(temp_dir, 'test_checkpoint.pth')
torch.save(checkpoint, model_path)

# Create arguments
opt = OptConfig()
model_new = SimpleModel()
optimizer_new = torch.optim.Adam(model_new.parameters(), lr=0.001)

ARGS = {
    'model': model_new,
    'model_path': model_path,
    'opt': opt,
    'optimizer': optimizer_new
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 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
parse_dynamic_params Ran
ran; our contract draft was wrong, not the code
qa276390/searchtrack/src/lib/model/decode.py
code served (permissive licence) · get_code("46f34dc9f9a94b67")
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 torch

torch.manual_seed(0)

# Analyze the function to understand parameter requirements:
# - params: 2D tensor, size(0) = num_insts, size(1) = sum(weight_nums) + sum(bias_nums)
# - channels: int, used for reshaping
# - weight_nums: list of ints, specifies weight dimensions for each layer
# - bias_nums: list of ints, specifies bias dimensions for each layer

# Set up small concrete values
num_insts = 2
channels = 4
num_layers = 2

# Define weight and bias numbers for 2 layers
# Layer 0 (not last): weight has some size, bias has some size
# Layer 1 (last): weight has some size, bias has some size
weight_nums = [12, 8]  # weight dimensions for each layer
bias_nums = [4, 1]     # bias dimensions for each layer

# params tensor must have:
# - dim 0: num_insts = 2
# - dim 1: sum(weight_nums) + sum(bias_nums) = 12 + 8 + 4 + 1 = 25
total_params = sum(weight_nums) + sum(bias_nums)
params = torch.randn(num_insts, total_params)

ARGS = {
    "params": params,
    "channels": channels,
    "weight_nums": weight_nums,
    "bias_nums": bias_nums,
}
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 — 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
queue Ran
ran on a synthesized input
hampen2929/CenterTrack/notebook/CenterTrack.py
code served (permissive licence) · get_code("53c06df9693423cb")
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 (3, 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 numpy as np

# The queue function takes two numpy arrays and:
# 1. vstacks them together
# 2. deletes the first row
# 3. returns the result
# This effectively shifts arr up and adds arr_add at the bottom

# Create small test arrays with distinct dimensions
arr = 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]])  # shape (3, 4)

arr_add = np.array([[13.0, 14.0, 15.0, 16.0]])  # shape (1, 4)

ARGS = {
    "arr": arr,
    "arr_add": arr_add
}
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 (3, 4) — shape (3, 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
DataParallel Not yet run
unverified (harvested, no recorded run)
xingyizhou/CenterTrack/src/lib/model/data_parallel.py
code served (permissive licence) · get_code("ddb5a5004c8da97a")
How it ran

no invocation recorded.

data_parallel Not yet run
unverified (harvested, no recorded run)
xingyizhou/CenterTrack/src/lib/model/data_parallel.py
code served (permissive licence) · get_code("61bf3906cbd2ee87")
How it ran

no invocation recorded.

generic_decode Not yet run
unverified (harvested, no recorded run)
xingyizhou/CenterTrack/src/lib/model/decode.py
code served (permissive licence) · get_code("7c992af205c8aa75")
How it ran

no invocation recorded.

load_model Not yet run
unverified (harvested, no recorded run)
qa276390/searchtrack/src/lib/model/model.py
code served (permissive licence) · get_code("b0dcae8d7c7f7659")
How it ran

no invocation recorded.

ltrb_amodal_decode Not yet run
unverified (harvested, no recorded run)
qa276390/searchtrack/src/lib/model/decode.py
code served (permissive licence) · get_code("581674d14c6feb31")
How it ran

no invocation recorded.

wh_decode Not yet run
unverified (harvested, no recorded run)
qa276390/searchtrack/src/lib/model/decode.py
code served (permissive licence) · get_code("a8979ecac7a82a7c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 8 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 5 the same month, 5 later) and 8 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
data_parallel 9 other papers
8 dated earlier, 1 the same month
1505.05424 (2015-05)
Weight Uncertainty in Neural Networks
1603.00831 (2016-03)
MOT16: A Benchmark for Multi-Object Tracking
1808.01244 (2018-08)
CornerNet: Detecting Objects as Paired Keypoints
and 6 more not named here
parse_dynamic_params 9 other papers
9 later
2201.00487 (2022-01)
Language as Queries for Referring Video Object Segmentation
2207.14012 (2022-07)
Video Mask Transfiner for High-Quality Video Instance Segmentation
aaai_25201 (2023 (year only))
InstanceFormer: An Online Video Instance Segmentation Framework
and 6 more not named here
compute_bin_loss 6 other papers
4 dated earlier, 1 the same month, 1 later
1603.00831 (2016-03)
MOT16: A Benchmark for Multi-Object Tracking
1904.07850 (2019-04)
Objects as Points
2001.03343 (2020-01)
RTM3D: Real-time Monocular 3D Detection from Object Keypoints for Autonomous Driving
and 3 more not named here
compute_res_loss 6 other papers
4 dated earlier, 1 the same month, 1 later
1603.00831 (2016-03)
MOT16: A Benchmark for Multi-Object Tracking
1904.07850 (2019-04)
Objects as Points
2001.03343 (2020-01)
RTM3D: Real-time Monocular 3D Detection from Object Keypoints for Autonomous Driving
and 3 more not named here
compute_rot_loss 6 other papers
4 dated earlier, 1 the same month, 1 later
1603.00831 (2016-03)
MOT16: A Benchmark for Multi-Object Tracking
1904.07850 (2019-04)
Objects as Points
2001.03343 (2020-01)
RTM3D: Real-time Monocular 3D Detection from Object Keypoints for Autonomous Driving
and 3 more not named here
DataParallel 5 other papers
4 dated earlier, 1 the same month
1603.00831 (2016-03)
MOT16: A Benchmark for Multi-Object Tracking
1904.07850 (2019-04)
Objects as Points
2001.04608 (2020-01)
Actions as Moving Points
and 2 more not named here
generic_decode 1 other paper
1 dated earlier
1603.00831 (2016-03)
MOT16: A Benchmark for Multi-Object Tracking
load_model 1 other paper
1 later
2111.01606 (2021-11)
PolyTrack: Tracking with Bounding Polygons

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 2026-11-21.

Repositories linked to this paper

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

Abstract

Tracking has traditionally been the art of following interest points through space and time. This changed with the rise of powerful deep networks. Nowadays, tracking is dominated by pipelines that perform object detection followed by temporal association, also known as tracking-by-detection. In this paper, we present a simultaneous detection and tracking algorithm that is simpler, faster, and more accurate than the state of the art. Our tracker, CenterTrack, applies a detection model to a pair of images and detections from the prior frame. Given this minimal input, CenterTrack localizes objects and predicts their associations with the previous frame. That's it. CenterTrack is simple, online (no peeking into the future), and real-time. It achieves 67.3% MOTA on the MOT17 challenge at 22 FPS and 89.4% MOTA on the KITTI tracking benchmark at 15 FPS, setting a new state of the art on both datasets. CenterTrack is easily extended to monocular 3D tracking by regressing additional 3D attributes. Using monocular video input, it achieves 28.3% AMOTA@0.2 on the newly released nuScenes 3D tracking benchmark, substantially outperforming the monocular baseline on this benchmark while running at 28 FPS.

For agents

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

get_harvested_code_for_paper("2004.01177")
get_code_for_paper("2004.01177")
have("2004.01177")

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

curl https://syntology.ai/api/ran/2004.01177.json

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

[![Syntology run record](https://syntology.ai/api/ran/2004.01177.svg)](https://syntology.ai/paper/2004.01177)

Connect an agent — have() is free.