SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.17359 · NeurIPS · 2023

SE(3) Diffusion Model-based Point Cloud Registration for Robust 6D Object Pose Estimation

Mathieu Salzmann, Jin Xie, Jian Yang, Haobo Jiang, Zheng Dang

arXiv · PDF · Open in the Atlas

Code that ran

14 of 14 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 1 has no contract verdict; 12 failures of our instrument; 0 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/2310.17359.json · badge: /api/ran/2310.17359.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
Jiang-HB/DiffusionReg canonical (classifier label)
classifier basis: census
13 ran · 0 no recorded run
jiang-hb/diffusionreg community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
DiffusionScheduler Ran
ran on a synthesized input · constructed an object; no result was computed
jiang-hb/diffusionreg/utils/diffusion_scheduler.py
code served (permissive licence) · get_code("8e98d0e10489ba04")
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 an instance of its own class, DiffusionScheduler: the run constructed the object and did not call it, so no result was computed. 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 numpy as np

torch.manual_seed(0)
np.random.seed(0)

class Options:
    def __init__(self):
        self.n_diff_steps = 10
        self.beta_1 = 0.0001
        self.beta_T = 0.02
        self.schedule_type = 'linear'
        self.S = 0.008

opts = Options()

ARGS = {
    'opts': opts
}
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 DiffusionScheduler — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a DiffusionScheduler 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
angle_difference Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Jiang-HB/DiffusionReg/modules/DCP/rpmnet_emb/pointnet_util.py
code served (permissive licence) · get_code("e6531f546e55f6af")
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, 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)

# src: [B, N, C] - batch of N source points with C channels
# dst: [B, M, C] - batch of M target points with C channels
# The function assumes l2-normalized unit vectors and computes angles between them

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

# Create l2-normalized unit vectors for src and dst
src = torch.randn(B, N, C)
src = src / torch.norm(src, dim=2, keepdim=True)  # normalize to unit length

dst = torch.randn(B, M, C)
dst = dst / torch.norm(dst, dim=2, keepdim=True)  # normalize to unit length

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, 3, 4) — shape (2, 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
compute_losses Ran
ran; our contract draft was wrong, not the code
Jiang-HB/DiffusionReg/utils/losses.py
code served (permissive licence) · get_code("9678a2e24b9ac1cb")
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

torch.manual_seed(0)

# opts is not used in the function body, so a simple dict works
opts = {}

# src_pcd should be a point cloud tensor
# Shape: (batch_size, num_points, 3) for 3D points
# Using batch_size=2, num_points=4, 3D coordinates
src_pcd = torch.randn(2, 4, 3)

# pred_transforms is a list of transformation matrices
# Each transform should be shape (batch_size, 4, 4) for 4x4 homogeneous matrices
# The function extracts R (3x3) and p (3x1) from each
pred_transforms = [
    torch.randn(2, 4, 4),
    torch.randn(2, 4, 4),
    torch.randn(2, 4, 4),
]

# transform_gt is a ground truth transformation matrix
# Shape: (batch_size, 4, 4)
transform_gt = torch.randn(2, 4, 4)

# loss_type can be 'mse' or 'mae'
loss_type = 'mae'

# reduction can be 'mean' or 'none'
reduction = 'mean'

ARGS = {
    'opts': opts,
    'src_pcd': src_pcd,
    'pred_transforms': pred_transforms,
    'transform_gt': transform_gt,
    'loss_type': loss_type,
    'reduction': reduction,
}
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
compute_losses_diff Ran
ran; our contract draft was wrong, not the code
Jiang-HB/DiffusionReg/utils/losses.py
code served (permissive licence) · get_code("d045a9d15eb48776")
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
torch.manual_seed(0)

# Build tiny tensors for point clouds and transforms
# Point clouds: (batch, num_points, 3)
batch_size = 2
num_points = 4
point_dim = 3

src_pcd0 = torch.randn(batch_size, num_points, point_dim)
src_pcd = torch.randn(batch_size, num_points, point_dim)

# Transforms: (batch, 4, 4) - homogeneous transformation matrices
# Format: [[R (3x3), p (3x1)], [0 0 0 1]]
transform_gt = torch.eye(4).unsqueeze(0).expand(batch_size, -1, -1).clone()
transform_gt[:, :3, 3] = torch.randn(batch_size, 3)  # add random translation

# Multiple predicted transforms (list of tensors)
num_iter = 3
pred_transforms = [
    torch.eye(4).unsqueeze(0).expand(batch_size, -1, -1).clone()
    for _ in range(num_iter)
]
for i, t in enumerate(pred_transforms):
    t[:, :3, 3] = torch.randn(batch_size, 3) + i * 0.1

# Simple opts object (not used in function body)
class Opts:
    pass

opts = Opts()

ARGS = {
    'opts': opts,
    'src_pcd0': src_pcd0,
    'src_pcd': src_pcd,
    'pred_transforms': pred_transforms,
    'transform_gt': transform_gt,
    'loss_type': 'mae',
    'reduction': 'mean'
}
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
evaluate_Rt Ran
ran, violated the contract we drafted
Jiang-HB/DiffusionReg/utils/criterion.py
code served (permissive licence) · get_code("94ee3d652c921dcd")
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

# Construct valid arguments for evaluate_Rt
# The function expects rotation matrices (3x3) and translation vectors

# Ground truth rotation matrix (3x3)
R_gt = np.array([
    [1.0, 0.0, 0.0],
    [0.0, 1.0, 0.0],
    [0.0, 0.0, 1.0]
], dtype=np.float64)

# Ground truth translation vector (can be any shape, will be flattened)
t_gt = np.array([0.1, 0.2, 0.3], dtype=np.float64)

# Predicted rotation matrix (3x3) - slightly different from ground truth
# Create a small rotation
angle = 0.05
R_pred = np.array([
    [np.cos(angle), -np.sin(angle), 0.0],
    [np.sin(angle), np.cos(angle), 0.0],
    [0.0, 0.0, 1.0]
], dtype=np.float64)

# Predicted translation vector
t_pred = np.array([0.12, 0.22, 0.32], dtype=np.float64)

ARGS = {
    "R_gt": R_gt,
    "t_gt": t_gt,
    "R_pred": R_pred,
    "t_pred": t_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, 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
get_postpool Ran
ran; our contract draft was wrong, not the code
Jiang-HB/DiffusionReg/modules/DCP/rpmnet_emb/feature_net.py
code served (permissive licence) · get_code("552ad03917659f63")
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 Sequential. 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)

ARGS = {
    "in_dim": 64,
    "out_dim": 32,
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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_prepool Ran
ran; our contract draft was wrong, not the code
Jiang-HB/DiffusionReg/modules/DCP/rpmnet_emb/feature_net.py
code served (permissive licence) · get_code("8778312931854a3b")
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 Sequential. 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)

ARGS = {
    "in_dim": 3,
    "out_dim": 64,
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
index_points Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Jiang-HB/DiffusionReg/modules/DCP/rpmnet_emb/pointnet_util.py
code served (permissive licence) · get_code("0188255c8db5f0c5")
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)

# points: [B, N, C] - batch size 2, N points 5, C channels 3
points = torch.randn(2, 5, 3)

# idx: [B, S] - batch size 2, S samples 4
# indices should be valid (0 to N-1, so 0 to 4)
idx = torch.tensor([[0, 2, 1, 3], [4, 1, 2, 0]], dtype=torch.long)

ARGS = {
    "points": points,
    "idx": idx
}
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
integrate_trans Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Jiang-HB/DiffusionReg/modules/DCP/dcp.py
code served (permissive licence) · get_code("d23093a32d8fdc84")
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, 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)

# Test case 1: batched torch tensors
R_batched_torch = torch.randn(2, 3, 3)
t_batched_torch = torch.randn(2, 3, 1)

# Test case 2: single torch tensors
R_single_torch = torch.randn(3, 3)
t_single_torch = torch.randn(3, 1)

# Test case 3: batched numpy arrays
R_batched_numpy = np.random.randn(2, 3, 3)
t_batched_numpy = np.random.randn(2, 3, 1)

# Test case 4: single numpy arrays
R_single_numpy = np.random.randn(3, 3)
t_single_numpy = np.random.randn(3, 1)

# Use batched torch case as the primary test
ARGS = {
    "R": R_batched_torch,
    "t": t_batched_torch,
}
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 3, dtype family float, shape (2, 4, 4) — shape (2, 4, 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
mAP Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Jiang-HB/DiffusionReg/utils/criterion.py
code served (permissive licence) · get_code("fa2f8f0b22fc04a6")
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

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

# Create batch size of 2, with 3x3 rotation matrices and 3D translation vectors
B = 2

# Generate valid rotation matrices (orthogonal matrices with det=1)
# Using QR decomposition to ensure orthogonality
def generate_rotation_matrix():
    A = np.random.randn(3, 3)
    Q, R = np.linalg.qr(A)
    # Ensure determinant is 1 (not -1)
    if np.linalg.det(Q) < 0:
        Q[:, 0] *= -1
    return Q

# Generate rotation matrices for predictions and ground truth
Rs_pred = np.array([generate_rotation_matrix() for _ in range(B)])
Rs_gt = np.array([generate_rotation_matrix() for _ in range(B)])

# Generate translation vectors (3D vectors)
ts_pred = np.random.randn(B, 3)
ts_gt = np.random.randn(B, 3)

ARGS = {
    "Rs_pred": Rs_pred,
    "ts_pred": ts_pred,
    "Rs_gt": Rs_gt,
    "ts_gt": ts_gt,
}
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 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
quat2mat Ran
ran; our contract draft was wrong, not the code
Jiang-HB/DiffusionReg/modules/DCP/dcp.py
code served (permissive licence) · get_code("838e2078ad9224da")
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 batch of quaternions in [x, y, z, w] format
# Shape: (batch_size, 4)
quat = torch.randn(2, 4)

ARGS = {
    "quat": quat
}
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
square_distance Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Jiang-HB/DiffusionReg/modules/DCP/rpmnet_emb/pointnet_util.py
pointer only (licence: MIT); the stored copy get_code serves, from fukexue/RGM, is pointer only (licence: NONE) · get_code("4db73dd57b5d6162")
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 fukexue/RGM, 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.

transform Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Jiang-HB/DiffusionReg/modules/DCP/dcp.py
code served (permissive licence) · get_code("d60988746155a232")
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, 5, 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
import numpy as np

torch.manual_seed(0)

# Test case 1: 3D batched tensors (torch)
pts_batched_torch = torch.randn(2, 5, 3)  # [bs=2, num_pts=5, 3]
trans_batched_torch = torch.randn(2, 4, 4)  # [bs=2, 4, 4]
# Make it a valid SE3 matrix (rotation + translation)
trans_batched_torch[:, :3, :3] = torch.nn.functional.normalize(torch.randn(2, 3, 3), dim=-1)
trans_batched_torch[:, 3, :] = torch.tensor([0., 0., 0., 1.])

ARGS = {
    "pts": pts_batched_torch,
    "trans": trans_batched_torch,
}
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 3, dtype family float, shape (2, 5, 3) — shape (2, 5, 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
transform Ran
ran; our contract draft was wrong, not the code
Jiang-HB/DiffusionReg/utils/losses.py
code served (permissive licence) · get_code("4e06974feca89b9d")
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 4, 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

torch.manual_seed(0)

# g is a transformation matrix with shape [..., 3, 4] (rotation + translation)
# The function extracts R from g[..., :3, :3] and p from g[..., :3, 3]
g = torch.randn(2, 3, 4)

# a is a set of points that will be transformed
# The function does torch.matmul(a, R.transpose(-1, -2)) + p[..., None, :]
# So a should have shape [..., N, 3] where the last dim is 3 to match R
a = torch.randn(2, 5, 3)

# normals is optional, shape should match a for broadcasting
# The function does: normals @ R.transpose(-1, -2)
normals = torch.randn(2, 5, 3)

ARGS = {
    "g": g,
    "a": a,
    "normals": normals
}
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, ndim 4, 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

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 8 are also present in other papers whose code we hold (8 with at least one paper dated earlier, 0 the same month, 3 later) and 6 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
index_points 42 other papers
26 dated earlier, 16 later · 6 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
2003.13479 (2020-03)
RPM-Net: Robust Point Matching using Learned Features
and 39 more not named here
integrate_trans 3 other papers
3 dated earlier · 1 only as an assembly brick
2103.05465 (2021-03)
PointDSC: Robust Point Cloud Registration using Deep Spatial Consistency
2203.14453 (2022-03)
SC 2 -PCR: A Second Order Spatial Compatibility for Efficient and Robust Point Cloud Registration
2305.10854 (2023-05)
3D Registration with Maximal Cliques
quat2mat 3 other papers
3 dated earlier
1806.01411 (2018-06)
FlowNet3D: Learning Scene Flow in 3D Point Clouds
2110.02210 (2021-10)
Mix3D: Out-of-Context Data Augmentation for 3D Scenes
2202.11292 (2022-02)
Reliable Inlier Evaluation for Unsupervised Point Cloud Registration
square_distance 3 other papers
2 dated earlier, 1 later
2003.13479 (2020-03)
RPM-Net: Robust Point Matching using Learned Features
2103.04256 (2021-03)
Robust Point Cloud Registration Framework Based on Deep Graph Matching
2604.05060 (2026-04)
R3PM-Net: Real-time, Robust, Real-world Point Matching Network
transform 3 other papers
3 dated earlier · 1 only as an assembly brick
2103.05465 (2021-03)
PointDSC: Robust Point Cloud Registration using Deep Spatial Consistency
aaai_19917 (2022 (year only))
DeTarNet: Decoupling Translation and Rotation by Siamese Network for Point Cloud Registration
2203.14453 (2022-03)
SC 2 -PCR: A Second Order Spatial Compatibility for Efficient and Robust Point Cloud Registration
angle_difference 2 other papers
1 dated earlier, 1 later
2003.13479 (2020-03)
RPM-Net: Robust Point Matching using Learned Features
2604.05060 (2026-04)
R3PM-Net: Real-time, Robust, Real-world Point Matching Network
get_postpool 1 other paper
1 dated earlier
2003.13479 (2020-03)
RPM-Net: Robust Point Matching using Learned Features
get_prepool 1 other paper
1 dated earlier
2003.13479 (2020-03)
RPM-Net: Robust Point Matching using Learned Features

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

In this paper, we introduce an SE(3) diffusion model-based point cloud registration framework for 6D object pose estimation in real-world scenarios. Our approach formulates the 3D registration task as a denoising diffusion process, which progressively refines the pose of the source point cloud to obtain a precise alignment with the model point cloud. Training our framework involves two operations: An SE(3) diffusion process and an SE(3) reverse process. The SE(3) diffusion process gradually perturbs the optimal rigid transformation of a pair of point clouds by continuously injecting noise (perturbation transformation). By contrast, the SE(3) reverse process focuses on learning a denoising network that refines the noisy transformation step-by-step, bringing it closer to the optimal transformation for accurate pose estimation. Unlike standard diffusion models used in linear Euclidean spaces, our diffusion model operates on the SE(3) manifold. This requires exploiting the linear Lie algebra se(3) associated with SE(3) to constrain the transformation transitions during the diffusion and reverse processes. Additionally, to effectively train our denoising network, we derive a registration-specific variational lower bound as the optimization objective for model learning. Furthermore, we show that our denoising network can be constructed with a surrogate registration model, making our approach applicable to different deep registration networks. Extensive experiments demonstrate that our diffusion registration framework presents outstanding pose estimation performance on the real-world TUD-L, LINEMOD, and Occluded-LINEMOD datasets. Code is available at https://github.com/Jiang-HB/DiffusionReg.

For agents

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

get_harvested_code_for_paper("2310.17359")
get_code_for_paper("2310.17359")
have("2310.17359")

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

curl https://syntology.ai/api/ran/2310.17359.json

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

[![Syntology run record](https://syntology.ai/api/ran/2310.17359.svg)](https://syntology.ai/paper/2310.17359)

Connect an agent — have() is free.