SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.12082 · CVPR · 2022

PlaneMVS: 3D Plane Reconstruction from Multi-View Stereo

Pan Ji, Yi Xu, Jiachen Liu, Xiaolei Huang, Qingan Yan, Changjiang Cai, Nitin Bansal

arXiv · PDF · Open in the Atlas

Code that ran

16 of 19 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 16 have no contract verdict; 0 failures of our instrument; 3 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/2203.12082.json · badge: /api/ran/2203.12082.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
oppo-us-research/planemvs canonical (classifier label)
classifier basis not recorded on the link
16 ran · 3 no recorded run
FunctionStatusWhere it lives
clean_segmentation Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/clean_seg.py
code served (permissive licence) · get_code("c511045350c22381")
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 cv2

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

# Image: (h, w, 3) RGB image
h, w = 480, 640
image = np.random.randint(30, 255, size=(h, w, 3)).astype(np.float32)

# Grids: (3, h, w) - coordinate grids for x, y, 1
grids = np.ones((3, h, w), dtype=np.float32)
grids[0] = np.tile(np.arange(w, dtype=np.float32), (h, 1))
grids[1] = np.tile(np.arange(h, dtype=np.float32)[:, None], (1, w))

# Planes: (num_planes, 4) - plane parameters [nx, ny, nz, d]
# Each plane is defined by normal (nx, ny, nz) and offset d
num_planes = 3
planes = np.array([
    [0.1, 0.2, 0.9, 5.0],
    [0.05, 0.15, 0.95, 6.0],
    [0.2, 0.1, 0.85, 4.0]
], dtype=np.float32)

# Masks: (num_planes, h, w) - binary masks for each plane
masks = np.zeros((num_planes, h, w), dtype=np.float32)
masks[0, 50:150, 100:300] = 1.0
masks[1, 200:350, 200:450] = 1.0
masks[2, 100:250, 350:550] = 1.0

# Plane IDs: (num_planes,) - unique identifiers for planes
plane_ids = np.array([1, 2, 3], dtype=np.int32)

# Ground truth depth: (h, w) - depth map
gt_depth = np.random.uniform(0.5, 10.0, size=(h, w)).astype(np.float32)
gt_depth[:30, :] = 0.0  # Mark some areas as invalid

ARGS = {
    'image': image,
    'grids': grids,
    'planes': planes,
    'masks': masks,
    'plane_ids': plane_ids,
    'gt_depth': gt_depth,
    'planeAreaThreshold': 200,
    'planeWidthThreshold': 10,
    'depthDiffThreshold': 0.1,
    'validAreaThreshold': 0.5,
    'brightThreshold': 20
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, 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
cluster Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/cluster.py
code served (permissive licence) · get_code("e69c3883b85c1033")
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 (2, 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 torch
from sklearn.cluster import KMeans

torch.manual_seed(0)

# The function expects:
# - camera_planes: array-like with shape (N, 3+) where first 3 columns are normals
# - cluster_anchor_num: int specifying number of clusters

# Create a small numpy array of camera planes
# Shape: (batch_size=4, features=4) where first 3 are normals
camera_planes = np.array([
    [0.1, 0.2, 0.3, 1.0],
    [0.4, 0.5, 0.6, 2.0],
    [0.7, 0.8, 0.9, 3.0],
    [0.2, 0.3, 0.4, 4.0],
], dtype=np.float32)

ARGS = {
    "camera_planes": camera_planes,
    "cluster_anchor_num": 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — 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 (2, 3) — shape (2, 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
get_all_7scenes_path Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_7scenes_data.py
code served (permissive licence) · get_code("e389daf1368730ee")
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 os
import tempfile
import shutil

# Create a minimal mock directory structure for the 7scenes dataset
temp_dir = tempfile.mkdtemp()

# Create directory structure with some color.png files
scenes_seqs = [
    ('chess', 'seq-01'),
    ('chess', 'seq-02'),
    ('chess', 'seq-04'),
    ('fire', 'seq-01'),
    ('office', 'seq-01'),
    ('office', 'seq-03'),
    ('pumpkin', 'seq-02'),
    ('stairs', 'seq-02'),
    ('stairs', 'seq-05'),
]

for scene, seq in scenes_seqs:
    scene_dir = os.path.join(temp_dir, scene, seq)
    os.makedirs(scene_dir, exist_ok=True)
    
    # Create dummy color.png files
    for i in range(2):
        color_file = os.path.join(scene_dir, f'frame_{i:05d}_color.png')
        with open(color_file, 'w') as f:
            f.write('dummy')

# Also add a test sequence that's not in train_seqs
test_scene_dir = os.path.join(temp_dir, 'chess', 'seq-03')
os.makedirs(test_scene_dir, exist_ok=True)
test_color_file = os.path.join(test_scene_dir, 'frame_00000_color.png')
with open(test_color_file, 'w') as f:
    f.write('dummy')

ARGS = {
    'data_root': temp_dir,
    'sample': -1,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tqdm — 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_all_tum_path Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_tum_data.py
code served (permissive licence) · get_code("9729d599ce2b3570")
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 os
import tempfile
import numpy as np

# Create a temporary directory structure that mimics TUM dataset layout
tmpdir = tempfile.mkdtemp()

# Create scene directories and files
scene_exts = ['freiburg1_desk', 'freiburg1_room', 'freiburg1_desk2', 'freiburg2_desk', 'freiburg3_long_office_household', 'freiburg3_long_office_household_validation']

for scene_ext in scene_exts:
    scene_name = 'rgbd_dataset_' + scene_ext
    scene_path = os.path.join(tmpdir, scene_name)
    os.makedirs(scene_path, exist_ok=True)
    
    # Create rgb directory and dummy png files
    rgb_dir = os.path.join(scene_path, 'rgb')
    os.makedirs(rgb_dir, exist_ok=True)
    for i in range(3):
        with open(os.path.join(rgb_dir, f'{i:010d}.png'), 'w') as f:
            f.write('dummy')
    
    # Create depth directory and dummy png files
    depth_dir = os.path.join(scene_path, 'depth')
    os.makedirs(depth_dir, exist_ok=True)
    for i in range(3):
        with open(os.path.join(depth_dir, f'{i:010d}.png'), 'w') as f:
            f.write('dummy')
    
    # Create rgb.txt with timestamps
    rgb_txt_path = os.path.join(scene_path, 'rgb.txt')
    rgb_timestamps = np.array([1.0, 2.0, 3.0])
    np.savetxt(rgb_txt_path, rgb_timestamps)
    
    # Create depth.txt with timestamps
    depth_txt_path = os.path.join(scene_path, 'depth.txt')
    depth_timestamps = np.array([1.1, 2.1, 3.1])
    np.savetxt(depth_txt_path, depth_timestamps)
    
    # Create groundtruth.txt with poses (timestamp + 7 values for position/quaternion)
    gt_txt_path = os.path.join(scene_path, 'groundtruth.txt')
    pose_data = np.column_stack([
        np.array([1.0, 2.0, 3.0]),  # timestamps
        np.random.randn(3, 7)        # 7 values per pose (3 position + 4 quaternion)
    ])
    np.savetxt(gt_txt_path, pose_data)

ARGS = {
    'data_root': tmpdir,
    'sample': -1
}
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_grid Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/clean_seg.py
code served (permissive licence) · get_code("a81a95d6d05862d4")
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 3, dtype family float, shape (3, 480, 640) 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

ARGS = {
    "h": 480,
    "w": 640,
}
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 3, dtype family float, shape (3, 480, 640) — shape (3, 480, 640) 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_rel_trans Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_7scenes_data.py
code served (permissive licence) · get_code("458c1ecc42ef30e6")
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

# Create two valid 4x4 homogeneous transformation matrices
# ref_pose: reference pose matrix
ref_pose = np.array([
    [1.0, 0.0, 0.0, 1.5],
    [0.0, 1.0, 0.0, 2.0],
    [0.0, 0.0, 1.0, 0.5],
    [0.0, 0.0, 0.0, 1.0]
], dtype=np.float64)

# src_pose: source pose matrix (must be invertible)
src_pose = np.array([
    [1.0, 0.0, 0.0, 0.8],
    [0.0, 1.0, 0.0, 1.2],
    [0.0, 0.0, 1.0, 0.3],
    [0.0, 0.0, 0.0, 1.0]
], dtype=np.float64)

ARGS = {
    "ref_pose": ref_pose,
    "src_pose": src_pose
}
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
kmeans_cluster Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/stereo_cluters.py
code served (permissive licence) · get_code("0704f949c1a27579")
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 (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
import numpy as np
from sklearn.cluster import KMeans

torch.manual_seed(0)

# Create small sample vectors for clustering
# Using shape (6, 4) - 6 samples, 4 dimensions
vectors = np.random.randn(6, 4).astype(np.float32)

# Number of clusters to find
cluster_anchor_num = 3

ARGS = {
    "vectors": vectors,
    "cluster_anchor_num": cluster_anchor_num,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — 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 (3, 4) — shape (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
load_color Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_pixel_planar.py
code served (permissive licence) · get_code("2015b55a787e9714")
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 3, dtype family int, shape (480, 640, 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 cv2
import numpy as np
import tempfile
import os

# Create a temporary directory and a dummy image file
temp_dir = tempfile.mkdtemp()
color_path = os.path.join(temp_dir, "test_image.jpg")

# Create a small dummy image (100x100 RGB)
dummy_image = np.random.randint(0, 256, (100, 100, 3), dtype=np.uint8)
cv2.imwrite(color_path, dummy_image)

ARGS = {
    "color_path": color_path,
    "h": 480,
    "w": 640,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (480, 640, 3) — shape (480, 640, 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
load_depth Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_pixel_planar.py
code served (permissive licence) · get_code("5a25f72413bb176b")
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 (480, 640) 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 cv2
import numpy as np
import tempfile
import os

# Create a temporary depth image file
temp_dir = tempfile.mkdtemp()
depth_path = os.path.join(temp_dir, "depth.png")

# Create a small depth image (16-bit grayscale) and save it
depth_image = np.random.randint(0, 65535, size=(480, 640), dtype=np.uint16)
cv2.imwrite(depth_path, depth_image)

ARGS = {
    "depth_path": depth_path,
    "divisor": 1000.0,
    "h": 480,
    "w": 640,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, 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 (480, 640) — shape (480, 640) 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
load_image Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/pre_process.py
code served (permissive licence) · get_code("fafa963ecd5d41cf")
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 3, dtype family int, shape (480, 640, 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 cv2
import numpy as np
import tempfile
import os

# Create a temporary image file for testing
temp_dir = tempfile.mkdtemp()
temp_image_path = os.path.join(temp_dir, "test_image.png")

# Create a simple test image (300x200 BGR image with some content)
test_image = np.random.randint(0, 256, (200, 300, 3), dtype=np.uint8)
cv2.imwrite(temp_image_path, test_image)

ARGS = {
    "image_path": temp_image_path,
    "resize": (640, 480)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (480, 640, 3) — shape (480, 640, 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
load_intrinsic Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_pixel_planar.py
code served (permissive licence) · get_code("b2225dd1ab5fd8a7")
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 (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 numpy as np
import tempfile
import os

# Create a temporary intrinsic matrix file
temp_dir = tempfile.mkdtemp()
intrinsic_path = os.path.join(temp_dir, "intrinsic.txt")

# Create a sample 4x4 intrinsic matrix (will be read as 3x3)
# Typical camera intrinsic matrix format
sample_intrinsic = np.array([
    [1296.0, 0.0, 648.0, 0.0],
    [0.0, 968.0, 484.0, 0.0],
    [0.0, 0.0, 1.0, 0.0],
    [0.0, 0.0, 0.0, 1.0]
])

np.savetxt(intrinsic_path, sample_intrinsic)

ARGS = {
    "intrinsic_path": intrinsic_path,
    "h": 480,
    "w": 640,
    "ori_h": 968,
    "ori_w": 1296
}
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 (3, 3) — shape (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
load_valid_ids Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/stereo_cluters.py
code served (permissive licence) · get_code("21c717dcb6058d30")
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 int, shape (4,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import cv2
import numpy as np
import tempfile
import os

# Create a temporary segmentation image for testing
# The function uses cv2.imread to read the file, so we need a real file
temp_dir = tempfile.mkdtemp()
seg_path = os.path.join(temp_dir, "test_seg.png")

# Create a simple segmentation map with some valid IDs and the invalid ID 65535
seg_data = np.array([
    [1, 2, 65535],
    [3, 4, 65535],
    [1, 2, 3]
], dtype=np.uint16)

# Write as PNG (cv2.imread with -1 flag reads all channels including alpha)
cv2.imwrite(seg_path, seg_data)

ARGS = {
    "seg_path": seg_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, 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 int, shape (4,) — shape (4,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
plane_to_depth Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/clean_seg.py
code served (permissive licence) · get_code("5ecc4b1ee40f1b3c")
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, 480, 640) 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)

# Analyze the function:
# - grids: reshaped to (3, -1), used in matrix multiplication with normals
#   normals @ grids means normals is (plane_num, 3) and grids is (3, h*w)
# - planes: indexed with [:, -1] and [:, :3], so shape (plane_num, 4+)
# - masks: multiplied with piece_depths which has shape (plane_num, h, w)
# - h, w: default 480, 640; used to reshape piece_depths

h = 480
w = 640
plane_num = 2

# grids should be (3, h, w) initially, then reshaped to (3, h*w)
grids = torch.randn(3, h, w)

# planes should be (plane_num, 4+) where last column is offset and first 3 are normals
# Using shape (plane_num, 4) for minimal valid planes
planes = torch.randn(plane_num, 4)

# masks should be (plane_num, h, w) to multiply element-wise with piece_depths
masks = torch.ones(plane_num, h, w)

ARGS = {
    'grids': grids,
    'planes': planes,
    'masks': masks,
    'h': h,
    'w': w
}
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, 480, 640) — shape (2, 480, 640) 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_files Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/split_tumrgbd.py
code served (permissive licence) · get_code("0ec4eab464808505")
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 tempfile
import os

# Create a temporary file with some test data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write("line 1\n")
temp_file.write("line 2\n")
temp_file.write("line 3\n")
temp_file.close()

stereo_file_path = temp_file.name

ARGS = {
    'stereo_file': stereo_file_path
}
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
read_pose Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_7scenes_data.py
code served (permissive licence) · get_code("58c3e75c2c18326e")
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 (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 numpy as np
import tempfile
import os

# Create a temporary pose file with sample data
temp_dir = tempfile.mkdtemp()
pose_file = os.path.join(temp_dir, "pose.txt")

# Write sample pose data (e.g., a 4x3 matrix representing poses)
sample_pose_data = np.array([
    [1.0, 0.0, 0.0, 5.0],
    [0.0, 1.0, 0.0, 6.0],
    [0.0, 0.0, 1.0, 7.0],
    [0.0, 0.0, 0.0, 1.0]
])
np.savetxt(pose_file, sample_pose_data)

ARGS = {
    "pose_path": pose_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 ndarray, ndim 2, dtype family float, shape (4, 4) — shape (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
read_rgb_mapping Ran
ran on a synthesized input
oppo-us-research/planemvs/data_preparation/generate_tum_data.py
code served (permissive licence) · get_code("f6157f7dfc05252f")
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 os
import tempfile
import numpy as np
from pathlib import Path

# Create a temporary directory structure to simulate the data
temp_dir = tempfile.mkdtemp()
data_root = temp_dir
scene = "test_scene"

# Create the scene directory and subdirectories
scene_dir = os.path.join(data_root, scene)
os.makedirs(scene_dir, exist_ok=True)
rgb_dir = os.path.join(scene_dir, 'rgb')
os.makedirs(rgb_dir, exist_ok=True)

# Create a mock rgb.txt file with timestamps
rgb_txt_path = os.path.join(scene_dir, 'rgb.txt')
timestamps = np.array([1.0, 2.0, 3.0, 4.0])
np.savetxt(rgb_txt_path, timestamps)

# Create mock RGB image files
for i in range(4):
    img_path = os.path.join(rgb_dir, f'frame_{i:06d}.png')
    Path(img_path).touch()

ARGS = {
    'data_root': data_root,
    'scene': scene
}
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
cluster_vis_3d_scatters Not yet run
unverified (harvested, no recorded run)
oppo-us-research/planemvs/data_preparation/stereo_cluters.py
code served (permissive licence) · get_code("4f1c557276c4cff2")
How it ran

no invocation recorded.

load_pose Not yet run
unverified (harvested, no recorded run)
oppo-us-research/planemvs/data_preparation/pre_process.py
code served (permissive licence) · get_code("85fd58a1c4771f86")
How it ran

no invocation recorded.

load_raw_seg Not yet run
unverified (harvested, no recorded run)
oppo-us-research/planemvs/data_preparation/pre_process.py
code served (permissive licence) · get_code("95474bdf353f674f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 19 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 19 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We present a novel framework named PlaneMVS for 3D plane reconstruction from multiple input views with known camera poses. Most previous learning-based plane reconstruction methods reconstruct 3D planes from single images, which highly rely on single-view regression and suffer from depth scale ambiguity. In contrast, we reconstruct 3D planes with a multi-view-stereo (MVS) pipeline that takes advantage of multi-view geometry. We decouple plane reconstruction into a semantic plane detection branch and a plane MVS branch. The semantic plane detection branch is based on a single-view plane detection framework but with differences. The plane MVS branch adopts a set of slanted plane hypotheses to replace conventional depth hypotheses to perform plane sweeping strategy and finally learns pixel-level plane parameters and its planar depth map. We present how the two branches are learned in a balanced way, and propose a soft-pooling loss to associate the outputs of the two branches and make them benefit from each other. Extensive experiments on various indoor datasets show that PlaneMVS significantly outperforms state-of-theart (SOTA) single-view plane reconstruction methods on both plane detection and 3D geometry metrics. Our method even outperforms a set of SOTA learning-based MVS methods thanks to the learned plane priors. To the best of our knowledge, this is the first work on 3D plane reconstruction within an end-to-end MVS framework.

For agents

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

get_harvested_code_for_paper("2203.12082")
get_code_for_paper("2203.12082")
have("2203.12082")

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

curl https://syntology.ai/api/ran/2203.12082.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.12082.svg)](https://syntology.ai/paper/2203.12082)

Connect an agent — have() is free.