SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.03833 · ECCV · 2022

Quasi-Balanced Self-Training on Noise-Aware Synthesis of Object Point Clouds for Closing Domain Gap

Kui Jia, Ke Chen, Zihao Wang, Yongwei Chen, Longkun Zou

arXiv · PDF · Open in the Atlas

Code that ran

13 of 15 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, 1 violated it and 10 have no contract verdict; 2 failures of our instrument; 2 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/2203.03833.json · badge: /api/ran/2203.03833.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
Gorilla-Lab-SCUT/QS3 canonical (classifier label)
classifier basis not recorded on the link
13 ran · 2 no recorded run
FunctionStatusWhere it lives
calc_loss Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/DefCls.py
code served (permissive licence) · get_code("1f8af716d9bd010b")
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 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 torch.nn as nn

torch.manual_seed(0)

# Create args object with DefCls_weight attribute
class Args:
    def __init__(self):
        self.DefCls_weight = 0.5

args = Args()

# Create logits dict with 'def_cls' key containing a tensor
# Shape: (batch_size=2, num_classes=4)
logits = {
    'def_cls': torch.randn(2, 4)
}

# Create labels tensor matching the batch size
# Shape: (batch_size=2,) with class indices
labels = torch.tensor([0, 2])

# Create a criterion (loss function)
# CrossEntropyLoss expects logits and class indices
criterion = nn.CrossEntropyLoss(reduction='none')

# Map all parameters to their values
ARGS = {
    'args': args,
    'logits': logits,
    'labels': labels,
    'criterion': criterion
}
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 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
calc_loss Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/PCM.py
code served (permissive licence) · get_code("f3c8ecae4d1fc6b1")
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 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 torch.nn as nn

torch.manual_seed(0)

# Create a small args object with cls_weight attribute
class Args:
    def __init__(self):
        self.cls_weight = 0.5

args = Args()

# Create logits dictionary with 'cls' key containing model predictions
# Batch size 2, number of classes 4
logits = {
    'cls': torch.randn(2, 4)
}

# Create mixup_vals: (Y_a, Y_b, lam)
# Y_a and Y_b are class labels (batch size 2)
# lam is mixing coefficient
Y_a = torch.tensor([0, 1])
Y_b = torch.tensor([2, 3])
lam = 0.7
mixup_vals = (Y_a, Y_b, lam)

# Create a loss criterion (CrossEntropyLoss expects logits and class indices)
criterion = nn.CrossEntropyLoss()

ARGS = {
    'args': args,
    'logits': logits,
    'mixup_vals': mixup_vals,
    'criterion': criterion
}
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 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
calc_loss Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/PosReg.py
code served (permissive licence) · get_code("6a987abc783d48a4")
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 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 torch.nn as nn

torch.manual_seed(0)

# Create a small args object with the required attribute
class Args:
    def __init__(self):
        self.PosReg_weight = 0.5

args = Args()

# Create logits dictionary with two classification outputs
# Batch size 2, 3 classes
logits = {
    'pos_cls1': torch.randn(2, 3),
    'pos_cls2': torch.randn(2, 3)
}

# Create pos_vals tuple with two label tensors and a lambda weight
pos_label_a = torch.randint(0, 3, (2,))
pos_label_b = torch.randint(0, 3, (2,))
lam = 0.7
pos_vals = (pos_label_a, pos_label_b, lam)

# Create a criterion (CrossEntropyLoss)
criterion = nn.CrossEntropyLoss()

# Define ARGS mapping all parameters
ARGS = {
    'args': args,
    'logits': logits,
    'pos_vals': pos_vals,
    'criterion': criterion
}
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 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
calculate_base_line Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/utils/pc_utils.py
code served (permissive licence) · get_code("db460975f051fdd4")
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 float64, ndim 0, 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

np.random.seed(0)

# Create rotation matrices (3x3) for left and right cameras
R1 = np.array([
    [0.9, -0.436, 0.0],
    [0.436, 0.9, 0.0],
    [0.0, 0.0, 1.0]
])

# Another rotation matrix for right camera
R2 = np.array([
    [0.9, 0.436, 0.0],
    [-0.436, 0.9, 0.0],
    [0.0, 0.0, 1.0]
])

# Translation vectors (3,) for left and right cameras
T1 = np.array([0.5, 0.2, 1.0])
T2 = np.array([0.3, 0.1, 0.9])

ARGS = {
    "R1": R1,
    "T1": T1,
    "R2": R2,
    "T2": T2,
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
chamfer_distance Ran
ran; our contract draft was wrong, not the code
Gorilla-Lab-SCUT/QS3/DefRec.py
code served (permissive licence) · get_code("f3aebb3c24fe89d6")
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 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)

# Construct valid arguments for chamfer_distance
# p1: size[B, N, C] - batch of point clouds
# p2: size[B, N, C] - batch of point clouds
# mask: size[B, N, C] - mask for valid points

B = 2  # batch size
N = 4  # number of points
C = 3  # coordinate dimensions

# Create point cloud tensors
p1 = torch.randn(B, N, C)
p2 = torch.randn(B, N, C)

# Create mask tensor - values should be 0 or 1
# The function uses mask[:, :, 0] so we need at least C >= 1
mask = torch.randint(0, 2, (B, N, C)).float()

ARGS = {
    "p1": p1,
    "p2": p2,
    "mask": mask,
}
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 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
get_K Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/utils/pc_utils.py
code served (permissive licence) · get_code("5ceb48e2cb69a51a")
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
import tempfile
import os

# Create a temporary directory and a temporary K matrix file
temp_dir = tempfile.mkdtemp()
K_file = os.path.join(temp_dir, "Camera_view_0_K.txt")

# Create a sample intrinsic matrix file
# Format: header line, then 3x3 matrix
# Standard intrinsic matrix format:
# fx  0  cx
#  0 fy  cy
#  0  0   1
K_matrix = np.array([
    [500.0, 0.0, 320.0],
    [0.0, 500.0, 240.0],
    [0.0, 0.0, 1.0]
])

# Write to file with a header line (skip_header=1 in the function)
with open(K_file, 'w') as f:
    f.write("Header line\n")
    np.savetxt(f, K_matrix, delimiter=' ')

ARGS = {
    "K_file": K_file
}
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_RT Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/utils/pc_utils.py
code served (permissive licence) · get_code("4002f116144fe495")
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
import tempfile
import os

# Create temporary directory for test files
tmpdir = tempfile.mkdtemp()

# Create sample R and T matrices as text files
# R should be a 3x3 rotation matrix
# T should be a 3x1 translation vector

# R1 file
R1_content = "# Rotation matrix 1\n0.9 0.1 0.0\n-0.1 0.9 0.0\n0.0 0.0 1.0"
R1_file = os.path.join(tmpdir, "Camera_view_0_R.txt")
with open(R1_file, 'w') as f:
    f.write(R1_content)

# T1 file
T1_content = "# Translation vector 1\n0.5\n0.2\n1.0"
T1_file = os.path.join(tmpdir, "Camera_view_0_T.txt")
with open(T1_file, 'w') as f:
    f.write(T1_content)

# R2 file
R2_content = "# Rotation matrix 2\n0.85 0.2 0.0\n-0.2 0.85 0.0\n0.0 0.0 1.0"
R2_file = os.path.join(tmpdir, "Camera_view_1_R.txt")
with open(R2_file, 'w') as f:
    f.write(R2_content)

# T2 file
T2_content = "# Translation vector 2\n1.0\n0.3\n0.8"
T2_file = os.path.join(tmpdir, "Camera_view_1_T.txt")
with open(T2_file, 'w') as f:
    f.write(T2_content)

ARGS = {
    "R1_file": R1_file,
    "T1_file": T1_file,
    "R2_file": R2_file,
    "T2_file": T2_file,
}
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
get_empty_dir Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/utils/scan_utils.py
code served (permissive licence) · get_code("6fafc92e5c20b3e0")
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, 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 os
import tempfile
import shutil

# Create temporary directories for testing
temp_dir = tempfile.mkdtemp()
data_path = os.path.join(temp_dir, 'data')
model_dir = os.path.join(temp_dir, 'models')

category = 'chair'
type_val = 'train'

# Set up directory structure
os.makedirs(os.path.join(data_path, category, type_val), exist_ok=True)
os.makedirs(os.path.join(model_dir, category, type_val), exist_ok=True)

# Create some dummy .obj files in the model directory
obj_names = ['model_1', 'model_2', 'model_3']
for obj_name in obj_names:
    obj_path = os.path.join(model_dir, category, type_val, f'{obj_name}.obj')
    with open(obj_path, 'w') as f:
        f.write('# dummy obj file\n')

# Create clean_pc.xyz for some models (to test the filtering logic)
os.makedirs(os.path.join(data_path, category, type_val, 'model_1'), exist_ok=True)
with open(os.path.join(data_path, category, type_val, 'model_1', 'clean_pc.xyz'), 'w') as f:
    f.write('0 0 0\n')

os.makedirs(os.path.join(data_path, category, type_val, 'model_2'), exist_ok=True)
# model_2 intentionally has no clean_pc.xyz

ARGS = {
    'data_path': data_path,
    'category': category,
    'type': type_val,
    'model_dir': model_dir
}
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 list, ndim 1, dtype family float — a list 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_obj_name Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/utils/scan_utils.py
code served (permissive licence) · get_code("6d8e4cbc2ae1db7e")
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 os
import tempfile
import shutil

# Create a temporary directory structure that mimics ModelNet
tmpdir = tempfile.mkdtemp()

# Create directory structure for a couple of categories
categories = ['chair', 'table']

for category in categories:
    # Handle the cabinet -> night_stand mapping
    dir_name = 'night_stand' if category.lower() == 'cabinet' else category.lower()
    
    train_dir = os.path.join(tmpdir, dir_name, 'train')
    test_dir = os.path.join(tmpdir, dir_name, 'test')
    
    os.makedirs(train_dir, exist_ok=True)
    os.makedirs(test_dir, exist_ok=True)
    
    # Create some dummy files with extensions
    for i in range(3):
        open(os.path.join(train_dir, f'obj_{i}.off'), 'w').close()
        open(os.path.join(test_dir, f'test_obj_{i}.off'), 'w').close()

# Prepare arguments
ARGS = {
    'modelnet_dir': tmpdir + '/',
    'category_list': ['chair', 'table']
}
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 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
knn Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Gorilla-Lab-SCUT/QS3/Models_Norm.py
code served (permissive licence) · get_code("cdd0141594039dcb")
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 Tensor, ndim 3, dtype family int, shape (2, 3, 2) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# x is used with:
# - transpose(2, 1): expects 3D tensor
# - matmul with transposed version: matrix multiplication
# - sum(x ** 2, dim=1, keepdim=True): sum over dimension 1
# - topk(k=k, dim=-1): get top k indices along last dimension
# 
# Shape should be (batch, features, num_points) based on the transpose operations
# Let's use batch=2, features=4, num_points=3

x = torch.randn(2, 4, 3)

# k is the number of nearest neighbors to find
# Should be less than num_points (3), so k=2
k = 2

ARGS = {
    "x": x,
    "k": k,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family int, shape (2, 3, 2) — shape (2, 3, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
l2_norm Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/Models_Norm.py
code served (permissive licence) · get_code("1869835ffa76f452")
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 a small input tensor (batch_size=2, features=4)
input_tensor = torch.randn(2, 4)

# The function has parameters: input and axit (note: likely typo for 'axis')
# axit is used as the dimension argument to torch.norm and torch.div
ARGS = {
    "input": input_tensor,
    "axit": 1
}
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
read_obj_file Ran
ran on a synthesized input
Gorilla-Lab-SCUT/QS3/utils/utils.py
code served (permissive licence) · get_code("1007ec1231eb8c3b")
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 tempfile
import os
import glob

# Create a temporary directory structure that mimics ModelNet
tmpdir = tempfile.mkdtemp()
modelnet_dir = tmpdir + "/"

# Create directory structure for test categories
categories = ["chair", "table"]
for category in categories:
    cat_lower = category.lower()
    train_dir = os.path.join(modelnet_dir, cat_lower, "train")
    test_dir = os.path.join(modelnet_dir, cat_lower, "test")
    os.makedirs(train_dir, exist_ok=True)
    os.makedirs(test_dir, exist_ok=True)
    
    # Create dummy .obj files
    for i in range(2):
        open(os.path.join(train_dir, f"model_{i}.obj"), "w").close()
        open(os.path.join(test_dir, f"model_{i}.obj"), "w").close()

# Also create night_stand directory (since cabinet maps to night_stand)
night_stand_train = os.path.join(modelnet_dir, "night_stand", "train")
night_stand_test = os.path.join(modelnet_dir, "night_stand", "test")
os.makedirs(night_stand_train, exist_ok=True)
os.makedirs(night_stand_test, exist_ok=True)
for i in range(2):
    open(os.path.join(night_stand_train, f"model_{i}.obj"), "w").close()
    open(os.path.join(night_stand_test, f"model_{i}.obj"), "w").close()

ARGS = {
    "modelnet_dir": modelnet_dir,
    "category_list": ["chair", "table", "cabinet"]
}
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 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
str2bool Ran
ran, violated the contract we drafted
Gorilla-Lab-SCUT/QS3/train_QBST_sim2real.py
pointer only (licence: MIT); the stored copy get_code serves, from miccaiif/WENO, is pointer only (licence: NONE) · get_code("d5be5e94e0feaeaf")
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 miccaiif/WENO, 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.

get_graph_feature Not yet run
unverified (harvested, no recorded run)
Gorilla-Lab-SCUT/QS3/Models_Norm.py
code served (permissive licence) · get_code("51e61f45df52e4b2")
How it ran

no invocation recorded.

reconstruction_loss Not yet run
unverified (harvested, no recorded run)
Gorilla-Lab-SCUT/QS3/DefRec.py
code served (permissive licence) · get_code("c79a55191d2c2289")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 2 the same month, 2 later) and 11 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
str2bool 175 other papers
46 dated earlier, 3 the same month, 124 later, 2 same year, order not determinable · 2 only as an assembly brick
1406.6247 (2014-06)
Recurrent Models of Visual Attention
1603.09382 (2016-03)
Deep Networks with Stochastic Depth
1710.10196 (2017-10)
Progressive Growing of GANs for Improved Quality, Stability, and Variation
and 172 more not named here
knn 52 other papers
16 dated earlier, 2 the same month, 34 later · 12 only as an assembly brick
1801.07829 (2018-01)
Dynamic Graph CNN for Learning on Point Clouds
1905.06292 (2019-05)
3D Point Cloud Generative Adversarial Network Based on Tree Structured Graph Convolutions
2005.06734 (2020-05)
Dense-Resolution Network for Point Cloud Classification and Segmentation
and 49 more not named here
chamfer_distance 1 other paper
1 dated earlier
2003.12641 (2020-03)
Self-Supervised Learning for Domain Adaptation on Point-Clouds
reconstruction_loss 1 other paper
1 dated earlier
2003.12641 (2020-03)
Self-Supervised Learning for Domain Adaptation on Point-Clouds

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

Semantic analyses of object point clouds are largely driven by releasing of benchmarking datasets, including synthetic ones whose instances are sampled from object CAD models. However, learning from synthetic data may not generalize to practical scenarios, where point clouds are typically incomplete, non-uniformly distributed, and noisy. Such a challenge of Simulation-to-Reality (Sim2Real) domain gap could be mitigated via learning algorithms of domain adaptation; however, we argue that generation of synthetic point clouds via more physically realistic rendering is a powerful alternative, as systematic non-uniform noise patterns can be captured. To this end, we propose an integrated scheme consisting of physically realistic synthesis of object point clouds via rendering stereo images via projection of speckle patterns onto CAD models and a novel quasi-balanced self-training designed for more balanced data distribution by sparsity-driven selection of pseudo labeled samples for long tailed classes. Experiment results can verify the effectiveness of our method as well as both of its modules for unsupervised domain adaptation on point cloud classification, achieving the state-ofthe-art performance. Source codes and the SpeckleNet synthetic dataset are available at https://github.com/Gorilla-Lab-SCUT/QS3.

For agents

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

get_harvested_code_for_paper("2203.03833")
get_code_for_paper("2203.03833")
have("2203.03833")

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

curl https://syntology.ai/api/ran/2203.03833.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.03833.svg)](https://syntology.ai/paper/2203.03833)

Connect an agent — have() is free.