SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.14505 · CVPR · 2023

Unsupervised Inference of Signed Distance Functions from Single Sparse Point Clouds without Learning Priors

Chao Chen, Yu-Shen Liu, Zhizhong Han

arXiv · PDF · Open in the Atlas

Code that ran

7 of 12 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, 1 violated it and 1 has no contract verdict; 4 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/2303.14505.json · badge: /api/ran/2303.14505.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
chenchao15/NeuralTPS canonical (classifier label)
classifier basis not recorded on the link
7 ran · 5 no recorded run
FunctionStatusWhere it lives
chamfer_distance_naive Ran
ran; our contract draft was wrong, not the code
chenchao15/NeuralTPS/im2mesh/common.py
code served (permissive licence) · get_code("8e6b5059f61d84e6")
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
import numpy as np

torch.manual_seed(0)

# The function expects two tensors of point clouds
# Shape: (batch_size, num_points, 3)
# The function uses .size(), .view(), .pow(), .sum(), .min(), .mean()
# So we need torch tensors, not numpy arrays

batch_size = 2
num_points = 4
point_dim = 3

points1 = torch.randn(batch_size, num_points, point_dim)
points2 = torch.randn(batch_size, num_points, point_dim)

ARGS = {
    "points1": points1,
    "points2": points2,
}
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 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
compute_iou Ran
ran, honoured the contract we drafted
chenchao15/NeuralTPS/im2mesh/common.py
code served (permissive licence) · get_code("0072de07b144b152")
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 (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 numpy as np
import torch

torch.manual_seed(0)

# Create two small occupancy tensors
# occ1: shape (2, 4) - batch of 2, flattened occupancy values of 4
# occ2: shape (2, 4) - same shape for comparison
occ1 = torch.rand(2, 4)
occ2 = torch.rand(2, 4)

ARGS = {
    "occ1": occ1,
    "occ2": occ2,
}
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 ndarray, 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
is_url Ran
ran, violated the contract we drafted
chenchao15/NeuralTPS/im2mesh/checkpoints.py
code served (permissive licence) · get_code("e39d881796e8d67e")
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 urllib

def is_url(url):
    scheme = urllib.parse.urlparse(url).scheme
    return scheme in ('http', 'https')

ARGS = {
    'url': 'https://example.com'
}
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 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
maxpool Ran
ran; our contract draft was wrong, not the code
chenchao15/NeuralTPS/im2mesh/encoder/pointnet.py
pointer only (licence: MIT); the stored copy get_code serves, from leolyliu/core4d-instructions, is pointer only (licence: NONE) · get_code("94c6fe25807ef01e")
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 leolyliu/core4d-instructions, 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.

near_point_idx Ran
ran on a synthesized input
chenchao15/NeuralTPS/sample_query_point.py
code served (permissive licence) · get_code("c1a1f5d9d78e39ad")
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 EagerTensor, ndim 1, dtype family int, shape (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 tensorflow as tf
import numpy as np

# Create small test tensors for the function
# The function expects two arrays of 3D points and counts of points

# array1: first point cloud with num_point1 points, each with 3 features
num_point1 = 2
array1_data = np.random.randn(num_point1, 3).astype(np.float32)
array1 = tf.constant(array1_data)

# array2: second point cloud with num_point2 points, each with 3 features
num_point2 = 3
array2_data = np.random.randn(num_point2, 3).astype(np.float32)
array2 = tf.constant(array2_data)

# Number of points in each array
num_point1_arg = num_point1
num_point2_arg = num_point2

# Number of features (default is 3 for 3D points)
num_features = 3

ARGS = {
    "array1": array1,
    "array2": array2,
    "num_point1": num_point1_arg,
    "num_point2": num_point2_arg,
    "num_features": num_features
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 1, dtype family int, shape (3,) — shape (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
safe_norm Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
chenchao15/NeuralTPS/sample_query_point.py
code served (permissive licence) · get_code("6043a99e59a7e14d")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

# Create a small TensorFlow tensor for testing
x = tf.constant([[1.0, 2.0, 3.0, 4.0],
                 [5.0, 6.0, 7.0, 8.0]], dtype=tf.float32)

# Define arguments for safe_norm function
ARGS = {
    'x': x,
    'epsilon': 1e-12,
    'axis': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
safe_norm_np Ran
ran; our contract draft was wrong, not the code
chenchao15/NeuralTPS/NeuralTPS.py
code served (permissive licence) · get_code("4a2d9b3a8039b465")
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 ndarray, 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 numpy as np

# Construct a small numpy array for testing
# Using shape (2, 4) - batch size 2, dimension 4
x = np.array([
    [1.0, 2.0, 3.0, 4.0],
    [5.0, 6.0, 7.0, 8.0]
], dtype=np.float32)

ARGS = {
    "x": x,
    "epsilon": 1e-12,
    "axis": 1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 (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
conv1d Not yet run
unverified (harvested, no recorded run)
chenchao15/NeuralTPS/tf_util.py
pointer only (licence: MIT); the stored copy get_code serves, from easyry/cra-pcn, is pointer only (licence: NONE) · get_code("7aab9e14068166d3")
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 easyry/cra-pcn, 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. No invocation recorded.

conv2d Not yet run
unverified (harvested, no recorded run)
chenchao15/NeuralTPS/tf_util.py
code served (permissive licence) · get_code("7ad640c844ad04d8")
How it ran

no invocation recorded.

conv2d_transpose Not yet run
unverified (harvested, no recorded run)
chenchao15/NeuralTPS/tf_util.py
pointer only (licence: MIT); the stored copy get_code serves, from easyry/cra-pcn, is pointer only (licence: NONE) · get_code("8632d18c12be28a9")
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 easyry/cra-pcn, 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. No invocation recorded.

get_2d_samples Not yet run
unverified (harvested, no recorded run)
chenchao15/NeuralTPS/p_encoder.py
code served (permissive licence) · get_code("7fb1ba5eeb5c7b42")
How it ran

no invocation recorded.

load_config Not yet run
unverified (harvested, no recorded run)
chenchao15/NeuralTPS/im2mesh/config.py
code served (permissive licence) · get_code("a2291d44f0411f42")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 10 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 1 the same month, 7 later) and 2 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
is_url 12 other papers
8 dated earlier, 4 later
1912.07372 (2019-12)
Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision
2001.05119 (2020-01)
Learning multiview 3D point cloud registration
2003.04618 (2020-03)
Convolutional Occupancy Networks
and 9 more not named here
compute_iou 9 other papers
6 dated earlier, 1 the same month, 2 later
2003.04618 (2020-03)
Convolutional Occupancy Networks
2104.01148 (2021-04)
Decomposing 3D Scenes into Objects via Unsupervised Volume Segmentation
2107.01899 (2021-07)
Ray-ONet: Efficient 3D Reconstruction From A Single RGB Image
and 6 more not named here
conv1d 8 other papers
5 dated earlier, 3 later
1803.05827 (2018-03)
Local Spectral Graph Convolution for Point Set Feature Learning
1809.07016 (2018-09)
Generating 3D Adversarial Point Clouds
2009.03137 (2020-09)
Towards Semantic Segmentation of Urban-Scale 3D Point Clouds: A Dataset, Benchmarks and Challenges
and 5 more not named here
chamfer_distance_naive 6 other papers
4 dated earlier, 2 later
2003.04618 (2020-03)
Convolutional Occupancy Networks
2107.01899 (2021-07)
Ray-ONet: Efficient 3D Reconstruction From A Single RGB Image
2204.01264 (2022-04)
Probabilistic Implicit Scene Completion
and 3 more not named here
conv2d_transpose 5 other papers
2 dated earlier, 3 later
1803.05827 (2018-03)
Local Spectral Graph Convolution for Point Set Feature Learning
1809.07016 (2018-09)
Generating 3D Adversarial Point Clouds
2308.13175 (2023-08)
GridPull: Towards Scalability in Learning Implicit Representations from 3D Point Clouds
and 2 more not named here
maxpool 4 other papers
2 dated earlier, 2 later
2104.12229 (2021-04)
Vector Neurons: A General Framework for SO(3)-Equivariant Networks
2203.16529 (2022-03)
CaDeX: Learning Canonical Deformation Coordinate Space for Dynamic Surface Representation via Neural Homeomorphism
2310.03740 (2023-10)
ContactGen: Generative Contact Modeling for Grasp Generation
and 1 more not named here
load_config 3 other papers
3 dated earlier
2003.04618 (2020-03)
Convolutional Occupancy Networks
2107.01899 (2021-07)
Ray-ONet: Efficient 3D Reconstruction From A Single RGB Image
2110.09415 (2021-10)
NeuralBlox: Real-Time Neural Representation Fusion for Robust Volumetric Mapping
conv2d 1 other paper
1 later
2308.13175 (2023-08)
GridPull: Towards Scalability in Learning Implicit Representations from 3D Point Clouds
safe_norm 1 other paper
1 dated earlier
2207.08631 (2022-07)
Latent Partition Implicit with Surface Codes for 3D Representation
safe_norm_np 1 other paper
1 dated earlier
2011.13495 (2020-11)
Neural-Pull: Learning Signed Distance Functions from Point Clouds by Learning to Pull Space onto Surfaces

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

It is vital to infer signed distance functions (SDFs) from 3D point clouds. The latest methods rely on generalizing the priors learned from large scale supervision. However, the learned priors do not generalize well to various geometric variations that are unseen during training, especially for extremely sparse point clouds. To resolve this issue, we present a neural network to directly infer SDFs from single sparse point clouds without using signed distance supervision, learned priors or even normals. Our insight here is to learn surface parameterization and SDFs inference in an end-to-end manner. To make up the sparsity, we leverage parameterized surfaces as a coarse surface sampler to provide many coarse surface estimations in training iterations, according to which we mine supervision and our thin plate splines (TPS) based network infers SDFs as smooth functions in a statistical way. Our method significantly improves the generalization ability and accuracy in unseen point clouds. Our experimental results show our advantages over the state-of-the-art methods in surface reconstruction for sparse point clouds under synthetic datasets and real scans.The code is available at https://github.com/chenchao15/NeuralTPS.

For agents

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

get_harvested_code_for_paper("2303.14505")
get_code_for_paper("2303.14505")
have("2303.14505")

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

curl https://syntology.ai/api/ran/2303.14505.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.14505.svg)](https://syntology.ai/paper/2303.14505)

Connect an agent — have() is free.