SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.17351 · CVPR · 2024

ICP-Flow: LiDAR Scene Flow Estimation with ICP

Holger Caesar, Yancong Lin

arXiv · PDF · Open in the Atlas

Code that ran

8 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; 1 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 5 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/2402.17351.json · badge: /api/ran/2402.17351.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
yanconglin/icp-flow canonical (classifier label)
classifier basis: census
8 ran · 1 no recorded run
FunctionStatusWhere it lives
average_meter Ran
ran, violated the contract we drafted
yanconglin/icp-flow/utils_eval.py
code served (permissive licence) · get_code("0bff81ed145208e2")
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 float, 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 numpy as np

torch.manual_seed(0)

# The function takes two parameters: errors and nums
# Both are iterables that get zipped together
# errors contains numeric values (error magnitudes)
# nums contains numeric values (counts/weights)
# The function multiplies each error by its corresponding num and sums them

# Create small test data
errors = [0.5, 0.3, 0.7]  # List of error values
nums = [10, 20, 15]  # List of counts/weights

ARGS = {
    "errors": errors,
    "nums": nums
}
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 float, 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
check_transformation Ran
ran; our contract draft was wrong, not the code
yanconglin/icp-flow/utils_check.py
code served (permissive licence) · get_code("e586d5d665de3dd8")
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 bool, 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

torch.manual_seed(0)

class ArgsConfig:
    def __init__(self):
        self.translation_frame = 0.5
        self.thres_iou = 0.3
        self.thres_rot = 0.2

args = ArgsConfig()
translation = torch.tensor([0.1, 0.2, 0.15])
rotation = torch.tensor([0.05, 0.08, 0.12, 0.9])
iou = 0.5

ARGS = {
    "args": args,
    "translation": translation,
    "rotation": rotation,
    "iou": iou
}
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 bool, 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
collate Ran
ran; our contract draft was wrong, not the code
yanconglin/icp-flow/dataset_argo.py
code served (permissive licence) · get_code("c749ef089e2a88e0")
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. 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)

# The collate function takes a batch parameter and returns it as-is
# A batch is typically a list of samples
# Each sample could be a tuple of (data, label) or just data

# Create a simple batch with a few samples
sample1 = (torch.randn(4, 3), torch.tensor(0))
sample2 = (torch.randn(4, 3), torch.tensor(1))
sample3 = (torch.randn(4, 3), torch.tensor(0))

batch = [sample1, sample2, sample3]

ARGS = {
    "batch": batch
}
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 list — a list 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
compute_epe_test Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
yanconglin/icp-flow/utils_eval.py
code served (permissive licence) · get_code("6caba5bd811fe4ac")
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 numpy as np

# Create small test flow arrays
# flow_pred: predicted flow vectors, shape (N, 3)
# flow_gt: ground truth flow vectors, shape (N, 3)

np.random.seed(0)

# Create 4 sample points with 3D flow vectors
num_points = 4
flow_pred = np.random.randn(num_points, 3).astype(np.float32)
flow_gt = np.random.randn(num_points, 3).astype(np.float32)

# Create mask with same shape as first dimension
mask = np.array([1, 1, 0, 1], dtype=np.float32)

ARGS = {
    "flow_pred": flow_pred,
    "flow_gt": flow_gt,
    "mask": mask,
}
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 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
crop_data Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
yanconglin/icp-flow/utils_eval.py
code served (permissive licence) · get_code("d9c2e38d130fd158")
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 numpy as np

class Args:
    def __init__(self):
        self.range_x = 5.0
        self.range_y = 4.0
        self.range_z = 0.5
        self.ground_slack = 0.2
        self.eval_ground = False

# Create sample data
num_points = 10
raw_points = np.random.randn(num_points, 3).astype(np.float32)
raw_points[:, 0] = np.random.uniform(-3, 3, num_points)  # x within range
raw_points[:, 1] = np.random.uniform(-2, 2, num_points)  # y within range
raw_points[:, 2] = np.random.uniform(1.0, 2.0, num_points)  # z above ground

time_indice = np.arange(num_points).astype(np.int32)
sd_labels = np.random.randint(0, 2, num_points).astype(np.int32)
fb_labels = np.random.randint(0, 2, num_points).astype(np.int32)
ego_motion_gt = np.eye(4, dtype=np.float32)
scene_flow = np.random.randn(num_points, 3).astype(np.float32)
data_path = "/dummy/path"

data = {
    'raw_points': raw_points,
    'time_indice': time_indice,
    'sd_labels': sd_labels,
    'fb_labels': fb_labels,
    'ego_motion_gt': ego_motion_gt,
    'scene_flow': scene_flow,
    'data_path': data_path
}

pred = np.random.randn(num_points, 2).astype(np.float32)

args = Args()

ARGS = {
    'args': args,
    'data': data,
    'pred': pred
}
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 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
nearest_neighbor Ran
ran, honoured the contract we drafted
yanconglin/icp-flow/utils_helper.py
code served (permissive licence) · get_code("909bc990df90a728")
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 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
from sklearn.neighbors import NearestNeighbors

# Set random seed for reproducibility
np.random.seed(0)

# Create small source points array (Nxm format)
# N=2 points, m=4 dimensions
src = np.random.rand(2, 4).astype(np.float32)

# Create small destination points array (Nxm format)
# N=3 points, m=4 dimensions (same as src)
dst = np.random.rand(3, 4).astype(np.float32)

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
numpy, sklearn — 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 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
segment_ground_thres Ran
ran; our contract draft was wrong, not the code
yanconglin/icp-flow/utils_ground.py
code served (permissive licence) · get_code("4d87bd7e643b9d38")
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 1, dtype family float, shape (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

class Args:
    def __init__(self):
        self.range_z = 10.0
        self.ground_slack = 0.5

points = np.array([
    [1.0, 2.0, 5.0],
    [3.0, 4.0, 15.0],
    [5.0, 6.0, 10.2],
    [7.0, 8.0, 11.0]
])

ARGS = {
    "args": 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 1, dtype family float, shape (4,) — shape (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
trackers2labels Ran
ran, violated the contract we drafted
yanconglin/icp-flow/utils_helper.py
code served (permissive licence) · get_code("2ac6524283b416eb")
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

# Construct arguments for trackers2labels function
# The function takes:
# - label_i: numpy array of labels
# - label_j: numpy array of labels
# - pairs: numpy array of shape (n_pairs, 2) containing pairs of labels to match

# Create label arrays with some positive labels (>= 0) and some negative (< 0)
label_i = np.array([0, 1, 0, 1, -1, -1], dtype=np.int32)
label_j = np.array([2, 3, 2, 3, -1, -1], dtype=np.int32)

# Create pairs array matching labels from label_i to label_j
# pair[0] must be in label_i, pair[1] must be in label_j
pairs = np.array([
    [0, 2],
    [1, 3]
], dtype=np.int32)

ARGS = {
    "label_i": label_i,
    "label_j": label_j,
    "pairs": pairs
}
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 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
segment_ground Not yet run
unverified (harvested, no recorded run)
yanconglin/icp-flow/utils_ground.py
code served (permissive licence) · get_code("83e559ee9409f049")
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, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 9 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.

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

Scene flow characterizes the 3D motion between two Li-DAR scans captured by an autonomous vehicle at nearby timesteps. Prevalent methods consider scene flow as pointwise unconstrained flow vectors that can be learned by either large-scale training beforehand or time-consuming optimization at inference. However, these methods do not take into account that objects in autonomous driving often move rigidly. We incorporate this rigid-motion assumption into our design, where the goal is to associate objects over scans and then estimate the locally rigid transformations. We propose ICP-Flow, a learning-free flow estimator. The core of our design is the conventional Iterative Closest Point (ICP) algorithm, which aligns the objects over time and outputs the corresponding rigid transformations. Crucially, to aid ICP, we propose a histogram-based initialization that discovers the most likely translation, thus providing a good starting point for ICP. The complete scene flow is then recovered from the rigid transformations. We outperform state-of-the-art baselines, including supervised models, on the Waymo dataset and perform competitively on Argoverse-v2 and nuScenes. Further, we train a feedforward neural network, supervised by the pseudo labels from our model, and achieve top performance among all models capable of real-time inference. We validate the advantage of our model on scene flow estimation with longer temporal gaps, up to 0.4 seconds where other models fail to deliver meaningful results.

For agents

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

get_harvested_code_for_paper("2402.17351")
get_code_for_paper("2402.17351")
have("2402.17351")

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

curl https://syntology.ai/api/ran/2402.17351.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.17351.svg)](https://syntology.ai/paper/2402.17351)

Connect an agent — have() is free.