SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2201.01501 · CVPR · 2022

Rethinking Depth Estimation for Multi-View Stereo: A Unified Representation

Zhenyu Wang, Ronggang Wang, Rui Peng, Yawen Lai, Rongjie Wang

arXiv · PDF · Open in the Atlas

Code that ran

12 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; 2 honoured the contract we drafted, 0 violated it and 7 have no contract verdict; 3 failures of our instrument; 3 no recorded run), measured 2026-11-21; 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/2201.01501.json · badge: /api/ran/2201.01501.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
prstrive/unimvsnet canonical (classifier label)
classifier basis not recorded on the link
12 ran · 3 no recorded run
FunctionStatusWhere it lives
classification_loss Ran
ran on a synthesized input
prstrive/unimvsnet/loss.py
code served (permissive licence) · get_code("d2cef21715760206")
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)

# prob_volume: probability volume tensor, shape (batch, depth_bins, height, width)
prob_volume = torch.rand(2, 4, 3, 3) + 0.1  # Add 0.1 to avoid near-zero values initially

# depth_values: depth values for each bin, shape (batch, depth_bins, height, width)
depth_values = torch.linspace(0.5, 10.0, 4).view(1, 4, 1, 1).expand(2, 4, 3, 3)

# interval: depth interval between bins, scalar or tensor
interval = torch.tensor(2.0)

# depth_gt: ground truth depth map, shape (batch, height, width)
depth_gt = torch.rand(2, 3, 3) * 8.0 + 1.0

# mask: binary mask for valid pixels, shape (batch, height, width)
mask = torch.ones(2, 3, 3, dtype=torch.bool)
mask[0, 0, 0] = False  # Make one pixel invalid

# weight: scalar weight for the loss
weight = 1.0

ARGS = {
    "prob_volume": prob_volume,
    "depth_values": depth_values,
    "interval": interval,
    "depth_gt": depth_gt,
    "mask": mask,
    "weight": weight,
}
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
depth_regression Ran
ran, honoured the contract we drafted
prstrive/unimvsnet/networks/module.py
pointer only (licence: MIT); the stored copy get_code serves, from truongkhang/cds-mvsnet, is pointer only (licence: NONE) · get_code("16a0f9b77eb60de7")
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 truongkhang/cds-mvsnet, 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.

homo_warping Ran
ran on a synthesized input
prstrive/unimvsnet/networks/module.py
code served (permissive licence) · get_code("d44ae99590b8f9bc")
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, 3, 8, 6) 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)

# Define the function parameters
batch_size = 2
channels = 4
height = 8
width = 6
num_depth = 3

# src_fea: source feature map with shape (batch, channels, height, width)
src_fea = torch.randn(batch_size, channels, height, width)

# src_proj: source projection matrix with shape (batch, 4, 4)
src_proj = torch.randn(batch_size, 4, 4)

# ref_proj: reference projection matrix with shape (batch, 4, 4)
ref_proj = torch.randn(batch_size, 4, 4)
# Make sure ref_proj is invertible by ensuring it's not singular
ref_proj = ref_proj + torch.eye(4).unsqueeze(0) * 5

# depth_values: depth values with shape (batch, num_depth)
depth_values = torch.linspace(0.5, 5.0, num_depth).unsqueeze(0).repeat(batch_size, 1)

ARGS = {
    "src_fea": src_fea,
    "src_proj": src_proj,
    "ref_proj": ref_proj,
    "depth_values": depth_values,
}
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, 3, 8, 6) — shape (2, 4, 3, 8, 6) 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
make_recursive_func Ran
ran; our contract draft was wrong, not the code
prstrive/unimvsnet/tools.py
code served (permissive licence) · get_code("b62bec09ceeaca79")
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 function. 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)

# Define a simple function to be made recursive
def example_func(x):
    """A simple function that doubles its input"""
    if isinstance(x, torch.Tensor):
        return x * 2
    elif isinstance(x, (int, float)):
        return x * 2
    else:
        return x

# The make_recursive_func takes a function as its only parameter
ARGS = {
    "func": example_func
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
motion_blur Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
prstrive/unimvsnet/datasets/blendedmvs.py
code served (permissive licence) · get_code("4e30ac617a07841b")
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 (4, 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 cv2
import numpy as np

# Create a small test image
# The function uses cv2.filter2D which expects a 2D or 3D numpy array
# Let's create a small color image (height=4, width=3, channels=3)
test_img = np.random.randint(0, 256, size=(4, 3, 3), dtype=np.uint8)

ARGS = {
    'img': test_img,
    'max_kernel_size': 3
}
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 (4, 3, 3) — shape (4, 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
mvs_loss Ran
ran on a synthesized input
prstrive/unimvsnet/loss.py
code served (permissive licence) · get_code("ba0d276529d1705e")
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.functional as F

torch.manual_seed(0)

# Small tensor dimensions
batch_size = 2
num_depth = 4
height = 3
width = 5

# Create depth ground truth for multiple stages
depth_gt_ms = {
    'stage1': torch.rand(batch_size, height, width) * 10.0 + 1.0,
    'stage2': torch.rand(batch_size, height, width) * 10.0 + 1.0,
    'stage3': torch.rand(batch_size, height, width) * 10.0 + 1.0,
}

# Create masks for multiple stages
mask_ms = {
    'stage1': torch.randint(0, 2, (batch_size, height, width)).float(),
    'stage2': torch.randint(0, 2, (batch_size, height, width)).float(),
    'stage3': torch.randint(0, 2, (batch_size, height, width)).float(),
}

# Create inputs for multiple stages
inputs = {
    'stage1': {
        'prob_volume': torch.sigmoid(torch.randn(batch_size, num_depth, height, width)),
        'depth': torch.rand(batch_size, height, width) * 10.0 + 1.0,
        'depth_values': torch.linspace(1.0, 11.0, num_depth).unsqueeze(0).unsqueeze(2).unsqueeze(3).expand(batch_size, num_depth, height, width),
        'interval': torch.tensor(1.0),
    },
    'stage2': {
        'prob_volume': torch.sigmoid(torch.randn(batch_size, num_depth, height, width)),
        'depth': torch.rand(batch_size, height, width) * 10.0 + 1.0,
        'depth_values': torch.linspace(1.0, 11.0, num_depth).unsqueeze(0).unsqueeze(2).unsqueeze(3).expand(batch_size, num_depth, height, width),
        'interval': torch.tensor(0.5),
    },
    'stage3': {
        'prob_volume': torch.sigmoid(torch.randn(batch_size, num_depth, height, width)),
        'depth': torch.rand(batch_size, height, width) * 10.0 + 1.0,
        'depth_values': torch.linspace(1.0, 11.0, num_depth).unsqueeze(0).unsqueeze(2).unsqueeze(3).expand(batch_size, num_depth, height, width),
        'interval': torch.tensor(0.25),
    },
}

ARGS = {
    'inputs': inputs,
    'depth_gt_ms': depth_gt_ms,
    'mask_ms': mask_ms,
    'mode': 'regression',
    'dlossw': [1.0, 0.8, 0.6],
}
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
read_camera_parameters Ran
ran on a synthesized input
prstrive/unimvsnet/filter/dypcd.py
code served (permissive licence) · get_code("0c21ad29e1868b8e")
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 a temporary file with valid camera parameters
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "camera_params.txt")

# Write camera parameters in the expected format
# Line 0: comment/header
# Lines 1-4: 4x4 extrinsics matrix
# Line 5: empty/comment
# Line 6: comment
# Lines 7-9: 3x3 intrinsics matrix

content = """# Camera parameters
1.0 0.0 0.0 0.1
0.0 1.0 0.0 0.2
0.0 0.0 1.0 0.3
0.0 0.0 0.0 1.0

# Intrinsics
520.0 0.0 320.0
0.0 520.0 240.0
0.0 0.0 1.0
"""

with open(temp_file, 'w') as f:
    f.write(content)

ARGS = {
    "filename": temp_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
read_img Ran
ran, honoured the contract we drafted
prstrive/unimvsnet/filter/dypcd.py
code served (permissive licence) · get_code("a1db096963f65ed2")
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 (4, 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
from PIL import Image
import tempfile
import os

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

# Create a small test image (4x3 pixels, RGB)
test_array = np.random.randint(0, 256, (4, 3, 3), dtype=np.uint8)
test_image = Image.fromarray(test_array, mode='RGB')
test_image.save(temp_filename)

ARGS = {
    "filename": temp_filename
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, 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 (4, 3, 3) — shape (4, 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
read_pfm Ran
ran; our contract draft was wrong, not the code
prstrive/unimvsnet/datasets/data_io.py
code served (permissive licence) · get_code("6b687b02c05fe850")
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 a temporary PFM file for testing
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test.pfm")

# Create a simple PFM file (color, 3-channel)
height, width = 2, 3
data = np.random.rand(height, width, 3).astype(np.float32)

# Write PFM file
with open(temp_file, 'wb') as f:
    # Header: PF for color
    f.write(b'PF\n')
    # Dimensions
    f.write(f'{width} {height}\n'.encode('utf-8'))
    # Scale (negative indicates little-endian)
    f.write(b'-1.0\n')
    # Data (flip vertically to match PFM format, write in little-endian)
    flipped_data = np.flipud(data)
    flipped_data.astype('<f').tofile(f)

ARGS = {
    'filename': temp_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
regression_loss Ran
ran on a synthesized input
prstrive/unimvsnet/loss.py
code served (permissive licence) · get_code("af922d39413f6095")
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.functional as F

torch.manual_seed(0)

# Create small tensors for depth estimation and ground truth
# Using distinct dimensions: batch=2, spatial=4, depth=3
batch_size = 2
spatial_dim = 4
depth_dim = 3

# depth_est: estimated depth map, shape (batch, depth_dim, spatial_dim)
depth_est = torch.randn(batch_size, depth_dim, spatial_dim)

# depth_gt: ground truth depth map, same shape as depth_est
depth_gt = torch.randn(batch_size, depth_dim, spatial_dim)

# mask: boolean mask to select valid pixels, same shape as depth_est
mask = torch.randint(0, 2, (batch_size, depth_dim, spatial_dim)).bool()

# weight: scalar weight to scale the loss
weight = torch.tensor(0.5)

ARGS = {
    'depth_est': depth_est,
    'depth_gt': depth_gt,
    'mask': mask,
    'weight': weight
}
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
tensor2numpy Ran
ran on a synthesized input
prstrive/unimvsnet/tools.py
code served (permissive licence) · get_code("e96100b6e68bc179")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The tensor2numpy function is decorated with make_recursive_func, which means
# it can accept nested structures (lists, tuples, dicts) or a single value.
# The function converts torch tensors to numpy arrays.

# Test with various input types that the function should handle:

# Single tensor
single_tensor = torch.randn(2, 4)

# Single numpy array
single_numpy = np.random.randn(2, 4)

# List of tensors and numpy arrays
list_input = [torch.randn(2, 4), np.random.randn(3, 5), torch.randn(1, 2)]

# Tuple of mixed types
tuple_input = (torch.randn(2, 4), np.random.randn(3, 5))

# Dict with nested structures
dict_input = {
    'tensor': torch.randn(2, 4),
    'numpy': np.random.randn(3, 5),
    'list': [torch.randn(1, 2), np.random.randn(2, 3)],
    'nested': {
        'deep_tensor': torch.randn(4, 3),
        'deep_numpy': np.random.randn(2, 2)
    }
}

# Complex nested structure
complex_input = {
    'tensors': [torch.randn(2, 4), torch.randn(3, 5)],
    'arrays': [np.random.randn(1, 2), np.random.randn(2, 3)],
    'mixed': (torch.randn(2, 4), np.random.randn(3, 5)),
    'tuple_of_lists': ([torch.randn(1, 1), np.random.randn(2, 2)], [torch.randn(3, 3)])
}

ARGS = {
    'vars': complex_input
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
winner_take_all Ran
ran on a synthesized input
prstrive/unimvsnet/networks/module.py
code served (permissive licence) · get_code("9fccefc4e39f9a9a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 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

torch.manual_seed(0)

# Create small tensors with distinct dimensions
batch_size = 2
depth_dim = 4
height = 3
width = 5

# prob_volume: (b, d, h, w) - probability volume
prob_volume = torch.rand(batch_size, depth_dim, height, width)

# depth_values: (b, d, h, w) - depth values corresponding to each probability
depth_values = torch.linspace(0, 10, depth_dim).view(1, depth_dim, 1, 1).expand(batch_size, depth_dim, height, width)

ARGS = {
    "prob_volume": prob_volume,
    "depth_values": depth_values,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 5) — shape (2, 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
read_gipuma_dmb Not yet run
unverified (harvested, no recorded run)
prstrive/unimvsnet/filter/gipuma.py
code served (permissive licence) · get_code("f6147a9b9d73ca5e")
How it ran

no invocation recorded.

read_pair_file Not yet run
unverified (harvested, no recorded run)
prstrive/unimvsnet/filter/dypcd.py
code served (permissive licence) · get_code("54c203150cdc9211")
How it ran

no invocation recorded.

save_images Not yet run
unverified (harvested, no recorded run)
prstrive/unimvsnet/tools.py
code served (permissive licence) · get_code("5c7bd35439896682")
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, 7 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 6 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
make_recursive_func 14 other papers
5 dated earlier, 8 later, 1 same year, order not determinable
1912.06378 (2019-12)
Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching
2007.10872 (2020-07)
Dense Hybrid Recurrent Multi-view Stereo Net with Dynamic Consistency Checking
2104.00681 (2021-04)
NeuralRecon: Real-Time Coherent 3D Reconstruction from Monocular Video
and 11 more not named here
read_pfm 12 other papers
3 dated earlier, 7 later, 2 same year, order not determinable
1912.06378 (2019-12)
Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching
2008.02268 (2020-08)
NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collections
2108.03824 (2021-08)
AA-RMVSNet: Adaptive Aggregation Recurrent Multi-view Stereo Network
and 9 more not named here
read_img 7 other papers
2 dated earlier, 4 later, 1 same year, order not determinable
2007.10872 (2020-07)
Dense Hybrid Recurrent Multi-view Stereo Net with Dynamic Consistency Checking
2108.03824 (2021-08)
AA-RMVSNet: Adaptive Aggregation Recurrent Multi-view Stereo Network
and 4 more not named here
depth_regression 5 other papers
2 dated earlier, 3 later
1912.06378 (2019-12)
Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching
2112.05999 (2021-12)
Curvature-guided dynamic scale networks for Multi-view Stereo
2208.02541 (2022-08)
MVSFormer: Multi-View Stereo by Learning Robust Image Features and Temperature-based Depth
and 2 more not named here
read_gipuma_dmb 5 other papers
4 dated earlier, 1 same year, order not determinable
1902.10556 (2019-02)
Recurrent MVSNet for High-resolution Multi-view Stereo Depth Inference
1905.02706 (2019-05)
Learning Unsupervised Multi-View Stereopsis via Robust Photometric Consistency
1912.06378 (2019-12)
Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching
and 2 more not named here
read_camera_parameters 4 other papers
2 dated earlier, 1 later, 1 same year, order not determinable
1912.06378 (2019-12)
Cascade Cost Volume for High-Resolution Multi-View Stereo and Stereo Matching
2104.13325 (2021-04)
MVS2D: Efficient Multi-view Stereo via Attention-Driven 2D Convolutions
and 1 more not named here
motion_blur 2 other papers
2 later
Zhang_GeoMVSNet_Learning_Multi-View_Stereo_With_Geometry_Perception_CVPR_2023_paper (2023 (year only))
GeoMVSNet: Learning Multi-View Stereo with Geometry Perception
2507.11333 (2025-07)
MonoMVSNet: Monocular Priors Guided Multi-View Stereo Network

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 2026-11-21.

Repositories linked to this paper

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

Abstract

Depth estimation is solved as a regression or classification problem in existing learning-based multi-view stereo methods. Although these two representations have recently demonstrated their excellent performance, they still have apparent shortcomings, e.g., regression methods tend to overfit due to the indirect learning cost volume, and classification methods cannot directly infer the exact depth due to its discrete prediction. In this paper, we propose a novel representation, termed Unification, to unify the advantages of regression and classification. It can directly constrain the cost volume like classification methods, but also realize the sub-pixel depth prediction like regression methods. To excavate the potential of unification, we design a new loss function named Unified Focal Loss, which is more uniform and reasonable to combat the challenge of sample imbalance. Combining these two unburdened modules, we present a coarse-to-fine framework, that we call UniMVSNet. The results of ranking first on both DTU and Tanks and Temples benchmarks verify that our model not only performs the best but also has the best generalization ability.

For agents

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

get_harvested_code_for_paper("2201.01501")
get_code_for_paper("2201.01501")
have("2201.01501")

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

curl https://syntology.ai/api/ran/2201.01501.json

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

[![Syntology run record](https://syntology.ai/api/ran/2201.01501.svg)](https://syntology.ai/paper/2201.01501)

Connect an agent — have() is free.