SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2207.09143 · ECCV · 2022

What Matters for 3D Scene Flow Network

Masayoshi Tomizuka, Wei Zhan, Zhe Liu, Yiyang Zhou, Guangming Wang, Hesheng Wang, Yunzhe Hu

arXiv · PDF · Open in the Atlas

Code that ran

4 of 9 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 2 have no contract verdict; 2 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/2207.09143.json · badge: /api/ran/2207.09143.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
IRMVLab/3DFlow canonical (classifier label)
classifier basis not recorded on the link
4 ran · 5 no recorded run
FunctionStatusWhere it lives
knn_point Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
IRMVLab/3DFlow/pointconv_util.py
code served (permissive licence) · get_code("6413b2b536ff6771")
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 int, shape (2, 3, 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)

# Define arguments for knn_point function
# nsample: max sample number in local region (int)
# xyz: all points, [B, N, C] (tensor)
# new_xyz: query points, [B, S, C] (tensor)

B = 2  # batch size
N = 5  # number of points in xyz
S = 3  # number of query points in new_xyz
C = 4  # coordinate dimension

nsample = 2  # max samples to retrieve

# Create random point clouds
xyz = torch.randn(B, N, C)  # [2, 5, 4]
new_xyz = torch.randn(B, S, C)  # [2, 3, 4]

ARGS = {
    "nsample": nsample,
    "xyz": xyz,
    "new_xyz": new_xyz,
}
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 int, shape (2, 3, 2) — shape (2, 3, 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
multiScaleLoss Ran
ran on a synthesized input
IRMVLab/3DFlow/models.py
code served (permissive licence) · get_code("6b42e89816dc50e9")
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 (1,) 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

torch.manual_seed(0)

# The function expects:
# - pred_flows: list of tensors with shape (batch, channels, points)
# - gt_flows: list of tensors with shape (batch, channels, points)
# - alpha: list of weights (default provided)

# Create multi-scale flow predictions and ground truth
# Each scale should have different spatial resolutions
num_scales = 4
batch_size = 2

pred_flows = [
    torch.randn(batch_size, 3, 16),  # scale 0: (batch=2, channels=3, points=16)
    torch.randn(batch_size, 3, 8),   # scale 1: (batch=2, channels=3, points=8)
    torch.randn(batch_size, 3, 4),   # scale 2: (batch=2, channels=3, points=4)
    torch.randn(batch_size, 3, 2),   # scale 3: (batch=2, channels=3, points=2)
]

gt_flows = [
    torch.randn(batch_size, 3, 16),  # scale 0: (batch=2, channels=3, points=16)
    torch.randn(batch_size, 3, 8),   # scale 1: (batch=2, channels=3, points=8)
    torch.randn(batch_size, 3, 4),   # scale 2: (batch=2, channels=3, points=4)
    torch.randn(batch_size, 3, 2),   # scale 3: (batch=2, channels=3, points=2)
]

alpha = [0.02, 0.04, 0.08, 0.16]

ARGS = {
    "pred_flows": pred_flows,
    "gt_flows": gt_flows,
    "alpha": alpha,
}
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 (1,) — shape (1,) 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
multiScaleLoss Ran
ran on a synthesized input
IRMVLab/3DFlow/models_occlusion.py
code served (permissive licence) · get_code("ba5d550ea86e437f")
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 (1,) 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

torch.manual_seed(0)

# The function expects:
# - pred_flows: list of tensors with shape (batch, channels, num_points)
# - gt_flows: list of tensors with shape (batch, channels, num_points)
# - mask: list of tensors with shape (batch, 1, num_points)
# - alpha: list of floats (default provided)

# Create multi-scale flow predictions and ground truth
# Using batch_size=2, different dimensions for clarity
num_scales = 4
batch_size = 2

pred_flows = []
gt_flows = []
masks = []

# Build 4 scales with decreasing spatial resolution
for scale_idx in range(num_scales):
    num_points = 16 - (scale_idx * 3)  # Decreasing: 16, 13, 10, 7
    channels = 3
    
    # pred_flows and gt_flows: (batch, channels, num_points)
    pred_flow = torch.randn(batch_size, channels, num_points)
    gt_flow = torch.randn(batch_size, channels, num_points)
    
    # mask: (batch, 1, num_points)
    mask = torch.ones(batch_size, 1, num_points)
    
    pred_flows.append(pred_flow)
    gt_flows.append(gt_flow)
    masks.append(mask)

alpha = [0.02, 0.04, 0.08, 0.16]

ARGS = {
    "pred_flows": pred_flows,
    "gt_flows": gt_flows,
    "mask": masks,
    "alpha": alpha
}
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 (1,) — shape (1,) 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
square_distance Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
IRMVLab/3DFlow/pointconv_util.py
code served (permissive licence) · get_code("60f32bcfa7d043b5")
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, 4, 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)

# Construct valid arguments for square_distance function
# Input:
#   src: source points, [B, N, C]
#   dst: target points, [B, M, C]
# Output:
#   dist: per-point square distance, [B, N, M]

B = 2  # batch size
N = 4  # number of source points
M = 3  # number of target points
C = 5  # coordinate dimension

src = torch.randn(B, N, C)
dst = torch.randn(B, M, C)

ARGS = {
    'src': src,
    'dst': dst,
}
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, 4, 3) — shape (2, 4, 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
evaluate_2d Not yet run
unverified (harvested, no recorded run)
IRMVLab/3DFlow/evaluation_utils.py
code served (permissive licence) · get_code("5aeab7a823d21f0f")
How it ran

no invocation recorded.

evaluate_3d Not yet run
unverified (harvested, no recorded run)
IRMVLab/3DFlow/evaluation_utils.py
code served (permissive licence) · get_code("cfb929a257ccd28f")
How it ran

no invocation recorded.

evaluate_3d_mask Not yet run
unverified (harvested, no recorded run)
IRMVLab/3DFlow/evaluation_utils.py
code served (permissive licence) · get_code("f05bbacf16ae9dd9")
How it ran

no invocation recorded.

postprocess Not yet run
unverified (harvested, no recorded run)
IRMVLab/3DFlow/cmd_args.py
code served (permissive licence) · get_code("fd8ada46834f05aa")
How it ran

no invocation recorded.

query_yes_no Not yet run
unverified (harvested, no recorded run)
IRMVLab/3DFlow/main_utils.py
code served (permissive licence) · get_code("ca593fc52a89c6a0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 1 the same month, 2 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
square_distance 75 other papers
17 dated earlier, 3 the same month, 54 later, 1 same year, order not determinable · 9 only as an assembly brick
1706.02413 (2017-06)
PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space
1811.07246 (2018-11)
PointConv: Deep Convolutional Networks on 3D Point Clouds
2003.00410 (2020-03)
PF-Net: Point Fractal Network for 3D Point Cloud Completion
and 72 more not named here
knn_point 21 other papers
4 dated earlier, 17 later · 1 only as an assembly brick
1612.00593 (2016-12)
PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation
2111.14819 (2021-11)
Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling
2202.07123 (2022-02)
Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework
and 18 more not named here

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

3D scene flow estimation from point clouds is a low-level 3D motion perception task in computer vision. Flow embedding is a commonly used technique in scene flow estimation, and it encodes the point motion between two consecutive frames. Thus, it is critical for the flow embeddings to capture the correct overall direction of the motion. However, previous works only search locally to determine a soft correspondence, ignoring the distant points that turn out to be the actual matching ones. In addition, the estimated correspondence is usually from the forward direction of the adjacent point clouds, and may not be consistent with the estimated correspondence acquired from the backward direction. To tackle these problems, we propose a novel all-to-all flow embedding layer with backward reliability validation during the initial scene flow estimation. Besides, we investigate and compare several design choices in key components of the 3D scene flow network, including the point similarity calculation, input elements of predictor, and predictor & refinement level design. After carefully choosing the most effective designs, we are able to present a model that achieves the stateof-the-art performance on FlyingThings3D and KITTI Scene Flow datasets. Our proposed model surpasses all existing methods by at least 38.2% on FlyingTh-ings3D dataset and 24.7% on KITTI Scene Flow dataset for EPE3D metric. We release our codes at https://github.com/IRMVLab/3DFlow.

For agents

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

get_harvested_code_for_paper("2207.09143")
get_code_for_paper("2207.09143")
have("2207.09143")

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

curl https://syntology.ai/api/ran/2207.09143.json

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

[![Syntology run record](https://syntology.ai/api/ran/2207.09143.svg)](https://syntology.ai/paper/2207.09143)

Connect an agent — have() is free.