SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2105.03211 · CVPR · 2021

NeRD: Neural 3D Reflection Symmetry Detector

Yi Ma, Yichao Zhou, Shichen Liu

arXiv · PDF · Open in the Atlas

Code that ran

10 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; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 10 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/2105.03211.json · badge: /api/ran/2105.03211.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
zhou13/nerd canonical (classifier label)
classifier basis: census
10 ran · 0 no recorded run
FunctionStatusWhere it lives
abn Ran
ran; our contract draft was wrong, not the code
zhou13/nerd/sym/models/hourglass_pose.py
code served (permissive licence) · get_code("a420bd26ffd79633")
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)

# The function `abn` takes a single parameter `planes` which is an integer
# representing the number of input/output channels for BatchNorm2d and ReLU.
# We'll use a small integer value.

planes = 64

ARGS = {
    "planes": planes
}
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
gold_spiral_sampling_patch Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zhou13/nerd/misc/find-radius.py
code served (permissive licence) · get_code("08fa3af6222de022")
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 (10, 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
import math

# Analyze the function signature and usage:
# - v: used with indexing v[0], v[1], v[2], and v[:, None], so needs to be a 1D numpy array of length 3
# - alpha: used in math.cos(alpha), so a scalar float
# - num_pts: used in range/indexing operations, so an integer

# Create the arguments
v = np.array([1.0, 0.5, 0.3])  # 3D direction vector
alpha = math.pi / 4  # angle in radians
num_pts = 10  # number of points to sample

ARGS = {
    "v": v,
    "alpha": alpha,
    "num_pts": num_pts
}
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 (10, 3) — shape (10, 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
l1_smooth Ran
ran; our contract draft was wrong, not the code
zhou13/nerd/sym/models/symmetry_net.py
code served (permissive licence) · get_code("c1f51d0f02b4e2c9")
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 2, dtype family float, shape (2, 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)

# Create small tensors for input and target
# The function expects two tensors of the same shape that support:
# - subtraction (-)
# - abs()
# - comparison (<)
# - torch.where()
input_tensor = torch.randn(2, 4)
target_tensor = torch.randn(2, 4)

# beta is a float parameter that controls the threshold
beta_value = 0.04

ARGS = {
    "input": input_tensor,
    "target": target_tensor,
    "beta": beta_value,
}
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 2, dtype family float, shape (2, 4) — shape (2, 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
orth Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zhou13/nerd/misc/find-radius.py
code served (permissive licence) · get_code("b6cf865210ac7f8c")
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 (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 'orth' takes a parameter 'v' which should be a sequence/array-like
# object with exactly 3 elements (x, y, z) based on the unpacking: x, y, z = v
# The function computes an orthogonal vector to v using numpy operations

v = np.array([1.0, 2.0, 3.0])

ARGS = {
    "v": v
}
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 (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
parmap Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zhou13/nerd/sym/parmapper.py
code served (permissive licence) · get_code("bc332be5812ad278")
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 generator. 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 multiprocessing as mp

torch.manual_seed(0)

# Simple test function that takes a single input and returns a result
def simple_func(x):
    return x * 2

# Another test function for kwargs
def func_with_kwargs(x, multiplier=1, offset=0):
    return x * multiplier + offset

# Function that uses *args
def func_with_args(x, *extra_args):
    result = x
    for arg in extra_args:
        result = result + arg
    return result

# Function that uses **kwargs unpacking
def func_with_kwstar(x, **kwargs):
    return x + kwargs.get('val', 0)

# Create a small sequence of inputs
seq = [1, 2, 3, 4, 5]

# Alternative: sequence of tuples for star=True
seq_tuples = [(1, 2), (3, 4), (5, 6)]

# Alternative: sequence of dicts for kwstar=True
seq_dicts = [{'x': 1, 'val': 10}, {'x': 2, 'val': 20}]

# Function for exception handling
def func_that_may_error(x):
    if x < 0:
        raise ValueError("Negative value")
    return x * 2

ARGS = {
    'fun': simple_func,
    'seq': seq,
    'N': 2,
    'Nt': 1,
    'chunksize': 1,
    'ordered': True,
    'daemon': False,
    'progress': False,
    'args': (),
    'kwargs': None,
    'star': False,
    'kwstar': False,
    'exception': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
IPython, ipywidgets, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type generator — a generator output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a generator 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
relu_dropout Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zhou13/nerd/sym/models/utils.py
code served (permissive licence) · get_code("e7a582d5f69b3f51")
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 2, dtype family float, shape (2, 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)

ARGS = {
    "x": torch.randn(2, 4),
    "p": 0.5,
    "inplace": False,
    "training": True,
}
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 2, dtype family float, shape (2, 4) — shape (2, 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
resample Ran
ran; our contract draft was wrong, not the code
zhou13/nerd/sym/models/utils.py
code served (permissive licence) · get_code("367253015238b2bf")
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 4, dtype family float, shape (2, 4, 6, 10) 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 torch.nn.functional as F

torch.manual_seed(0)

# Create a small 4D input tensor (batch=2, channels=4, height=3, width=5)
# F.interpolate expects 4D or 5D tensors for spatial operations
x = torch.randn(2, 4, 3, 5)

# scale_factor to upsample the spatial dimensions
scale_factor = 2.0

ARGS = {
    'x': x,
    'scale_factor': scale_factor
}
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 4, dtype family float, shape (2, 4, 6, 10) — shape (2, 4, 6, 10) 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
sample_sphere Ran
ran; our contract draft was wrong, not the code
zhou13/nerd/sym/datasets.py
code served (permissive licence) · get_code("7cce33ddd68a3664")
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 (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 math
import random
import numpy as np
import numpy.linalg as LA

# Set seed for reproducibility
random.seed(42)
np.random.seed(42)

# Create arguments for sample_sphere
v = np.array([1.0, 0.0, 0.0])  # A unit vector in 3D space
theta0 = math.pi / 6  # 30 degrees
theta1 = math.pi / 3  # 60 degrees

ARGS = {
    "v": v,
    "theta0": theta0,
    "theta1": theta1,
}
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 (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
sample_symmetry Ran
ran; our contract draft was wrong, not the code
zhou13/nerd/sym/datasets.py
code served (permissive licence) · get_code("e04758e29ed93fad")
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 (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 math
import random
import numpy as np
import numpy.linalg as LA

# Set seed for reproducibility
random.seed(42)
np.random.seed(42)

# Define arguments for sample_symmetry
w0 = np.array([1.0, 2.0, 3.0])  # 3D vector
theta0 = math.pi / 6  # 30 degrees
theta1 = math.pi / 4  # 45 degrees
delta = 1  # scalar multiplier

ARGS = {
    "w0": w0,
    "theta0": theta0,
    "theta1": theta1,
    "delta": delta
}
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 (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
warp Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zhou13/nerd/sym/models/utils.py
code served (permissive licence) · get_code("34ac6f4ec368f69a")
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 5, dtype family float, shape (2, 4, 6, 3, 5) 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 torch.nn.functional as F

torch.manual_seed(0)

# x0: [N, C, H, W] - feature maps
# N=2 (batch), C=4 (channels), H=3 (height), W=5 (width)
x0 = torch.randn(2, 4, 3, 5)

# S: [N, 4, 4] - transformation matrix
# N=2 (batch), 4x4 projection/transformation matrices
S = torch.randn(2, 4, 4)

# gamma: [D] - depth values
# D=6 (depth dimension)
gamma = torch.randn(6)

ARGS = {
    "x0": x0,
    "S": S,
    "gamma": gamma,
}
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 5, dtype family float, shape (2, 4, 6, 3, 5) — shape (2, 4, 6, 3, 5) 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

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, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 2 later) and 8 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
gold_spiral_sampling_patch 2 other papers
1 dated earlier, 1 later
1910.06316 (2019-10)
NeurVPS: Neural Vanishing Point Scanning via Conic Convolution
2203.08586 (2022-03)
Deep vanishing point detection: Geometric priors make dataset variations vanish
orth 2 other papers
1 dated earlier, 1 later
1910.06316 (2019-10)
NeurVPS: Neural Vanishing Point Scanning via Conic Convolution
2203.08586 (2022-03)
Deep vanishing point detection: Geometric priors make dataset variations vanish

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

Recent advances have shown that symmetry, a structural prior that most objects exhibit, can support a variety of single-view 3D understanding tasks. However, detecting 3D symmetry from an image remains a challenging task. Previous works either assume the symmetry is given or detect the symmetry with a heuristic-based method. In this paper, we present NeRD, a Neural 3D Reflection Symmetry Detector, which combines the strength of learning-based recognition and geometry-based reconstruction to accurately recover the normal direction of objects' mirror planes. Specifically, we enumerate the symmetry planes with a coarse-to-fine strategy and find the best ones by building 3D cost volumes to examine the intra-image pixel correspondence from the symmetry. Our experiments show that the symmetry planes detected with our method are significantly more accurate than the planes from direct CNN regression on both synthetic and real datasets. More importantly, we also demonstrate that the detected symmetry can be used to improve the performance of downstream tasks such as pose estimation and depth map regression by a wide margin over existing methods. The code of this paper has been made public at https://github.com/zhou13/nerd.

For agents

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

get_harvested_code_for_paper("2105.03211")
get_code_for_paper("2105.03211")
have("2105.03211")

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

curl https://syntology.ai/api/ran/2105.03211.json

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

[![Syntology run record](https://syntology.ai/api/ran/2105.03211.svg)](https://syntology.ai/paper/2105.03211)

Connect an agent — have() is free.