SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.18146 · CVPR · 2024

3DSFLabelling: Boosting 3D Scene Flow Estimation by Pseudo Auto-labelling

Guangming Wang, Jiuming Liu, Hesheng Wang, Yi Shan, Chaokang Jiang, Phigent Robotics, Dalong Du, Zhujin Liang, Zhuang Ma, Zhenqiang Liu

arXiv · PDF · Open in the Atlas

Code that ran

9 of 10 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; 3 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 6 failures of our instrument; 1 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/2402.18146.json · badge: /api/ran/2402.18146.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
jiangchaokang/3dsflabelling canonical (classifier label)
classifier basis: census
9 ran · 1 no recorded run
FunctionStatusWhere it lives
flip_point_cloud Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jiangchaokang/3dsflabelling/sf_model/FLOT/flot/datasets/augmentation.py
pointer only (licence: MIT); the stored copy get_code serves, from zhangyushan3/diffsf, is pointer only (licence: NONE) · get_code("21e4495ad40554c8")
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 zhangyushan3/diffsf, 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.

flip_scene_flow Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jiangchaokang/3dsflabelling/sf_model/FLOT/flot/datasets/augmentation.py
pointer only (licence: MIT); the stored copy get_code serves, from zhangyushan3/diffsf, is pointer only (licence: NONE) · get_code("c1a1a3e037cf14ba")
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 zhangyushan3/diffsf, 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.

normal_frame Ran
ran, honoured the contract we drafted
jiangchaokang/3dsflabelling/Gen_SF_label/lidarkitti.py
code served (permissive licence) · get_code("5de0939d35433a36")
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

# The function takes 'points' as input and performs array indexing and concatenation
# It expects a numpy array with shape (..., 3) where the last dimension has at least 3 elements
# The function extracts x, y, z coordinates and reorders them to y, x, z

# Create a small sample points array
# Shape: (2, 3) - batch of 2 points with x, y, z coordinates
points = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0]
], dtype=np.float32)

ARGS = {
    "points": points
}
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
normal_frame_nusc Ran
ran, honoured the contract we drafted
jiangchaokang/3dsflabelling/Gen_SF_label/lidarkitti.py
code served (permissive licence) · get_code("c85abdb8e140b0f7")
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

# The function takes a points array and performs coordinate transformations
# Points should be a numpy array with shape (..., 3) where the last dimension has at least 3 elements
# The function indexes: points[..., 0], points[..., 1], points[..., 2]

# Create a small sample points array with shape (2, 3)
# This represents 2 points in 3D space
points = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0]
], dtype=np.float32)

ARGS = {
    "points": points
}
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
random_flip_pc Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jiangchaokang/3dsflabelling/sf_model/FLOT/flot/datasets/augmentation.py
pointer only (licence: MIT); the stored copy get_code serves, from zhangyushan3/diffsf, is pointer only (licence: NONE) · get_code("a927fe2691f83c7f")
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 zhangyushan3/diffsf, 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.

rot_normal_frame Ran
ran, honoured the contract we drafted
jiangchaokang/3dsflabelling/Gen_SF_label/lidarkitti.py
code served (permissive licence) · get_code("09e3b16c3a6b283f")
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, 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

# Create a valid 3x3 rotation matrix
# Using a simple rotation matrix around z-axis by 45 degrees
angle = np.pi / 4
R = np.array([
    [np.cos(angle), -np.sin(angle), 0],
    [np.sin(angle), np.cos(angle), 0],
    [0, 0, 1]
])

ARGS = {
    "R": R
}
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, 3) — shape (3, 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
so3_relative_angle Ran
ran; our contract draft was wrong, not the code
jiangchaokang/3dsflabelling/Gen_SF_label/rsf_utils.py
code served (permissive licence) · get_code("a6eb1c968553d38d")
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

torch.manual_seed(0)

# Create two batches of valid rotation matrices
# A valid rotation matrix has trace in [-1, 3] and determinant 1
# For simplicity, we'll create small rotation matrices

# Identity matrix is a valid rotation matrix with trace = 3
identity_3x3 = torch.eye(3)

# Create a simple rotation around z-axis by 45 degrees
angle = torch.tensor(3.14159 / 4)  # 45 degrees
cos_a = torch.cos(angle)
sin_a = torch.sin(angle)
rot_z = torch.tensor([
    [cos_a, -sin_a, 0.0],
    [sin_a, cos_a, 0.0],
    [0.0, 0.0, 1.0]
])

# Batch of 2 rotation matrices
R1 = torch.stack([identity_3x3, rot_z], dim=0)
R2 = torch.stack([identity_3x3, identity_3x3], dim=0)

ARGS = {
    "R1": R1,
    "R2": R2,
    "eps": 0.0001
}
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 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
so3_rotation_angle Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jiangchaokang/3dsflabelling/Gen_SF_label/rsf_utils.py
code served (permissive licence) · get_code("537464a70617bdd8")
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

torch.manual_seed(0)

# Create a batch of valid rotation matrices
# For a rotation matrix, trace should be in range [-1, 3]
# Identity matrix has trace 3
# A 180-degree rotation around z-axis has trace -1

# Create batch of 2 rotation matrices (3x3 each)
identity = torch.eye(3)

# 180-degree rotation around z-axis
rotation_180_z = torch.tensor([
    [-1.0, 0.0, 0.0],
    [0.0, -1.0, 0.0],
    [0.0, 0.0, 1.0]
])

# Stack into batch of shape (2, 3, 3)
R = torch.stack([identity, rotation_180_z])

ARGS = {
    "R": R,
    "eps": 0.0001
}
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 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
symmetric_orthogonalization Ran
ran; our contract draft was wrong, not the code
jiangchaokang/3dsflabelling/Gen_SF_label/rsf_utils.py
code served (permissive licence) · get_code("9ea0cd6b589df9ea")
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 3, dtype family float, shape (2, 3, 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 torch

torch.manual_seed(0)

# Create a small batch of 9D vectors to be reshaped into 3x3 matrices
batch_size = 2
x = torch.randn(batch_size, 9)

# Parameter d specifies the dimension of the square matrix (3x3 in this case)
d = 3

ARGS = {
    "x": x,
    "d": d,
}
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 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 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
no_detection_return Not yet run
unverified (harvested, no recorded run)
jiangchaokang/3dsflabelling/Gen_SF_label/inference.py
code served (permissive licence) · get_code("e17127edd3f23eb4")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 3 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 3 later) and 7 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
flip_point_cloud 1 other paper
1 later
2403.05327 (2024-03)
DiffSF: Diffusion Models for Scene Flow Estimation
flip_scene_flow 1 other paper
1 later
2403.05327 (2024-03)
DiffSF: Diffusion Models for Scene Flow Estimation
random_flip_pc 1 other paper
1 later
2403.05327 (2024-03)
DiffSF: Diffusion Models for Scene Flow Estimation

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

Learning 3D scene flow from LiDAR point clouds presents significant difficulties, including poor generalization from synthetic datasets to real scenes, scarcity of realworld 3D labels, and poor performance on real sparse Li-DAR point clouds. We present a novel approach from the perspective of auto-labelling, aiming to generate a large number of 3D scene flow pseudo labels for real-world Li-DAR point clouds. Specifically, we employ the assumption of rigid body motion to simulate potential object-level rigid movements in autonomous driving scenarios. By updating different motion attributes for multiple anchor boxes, the rigid motion decomposition is obtained for the whole scene. Furthermore, we developed a novel 3D scene flow data augmentation method for global and local motion. By perfectly synthesizing target point clouds based on augmented motion parameters, we easily obtain lots of 3D scene flow labels in point clouds highly consistent with real scenarios. On multiple real-world datasets including LiDAR KITTI, nuScenes, and Argoverse, our method outperforms all previous supervised and unsupervised methods without requiring manual labelling. Impressively, our method achieves a tenfold reduction in EPE3D metric on the LiDAR KITTI dataset, reducing it from 0.190m to a mere 0.008m error.

For agents

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

get_harvested_code_for_paper("2402.18146")
get_code_for_paper("2402.18146")
have("2402.18146")

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

curl https://syntology.ai/api/ran/2402.18146.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.18146.svg)](https://syntology.ai/paper/2402.18146)

Connect an agent — have() is free.