SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2003.08348 · 2020

Multi-View Optimization of Local Feature Geometry

arXiv · PDF · Open in the Atlas

Code that ran

7 of 9 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 7 have no contract verdict; 0 failures of our instrument; 2 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/2003.08348.json · badge: /api/ran/2003.08348.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
mihaidusmanu/local-feature-refinement canonical (classifier label)
classifier basis: census
7 ran · 2 no recorded run
FunctionStatusWhere it lives
extract_patches Ran
ran on a synthesized input
mihaidusmanu/local-feature-refinement/two-view-refinement/refinement.py
code served (permissive licence) · get_code("e2a8ccc8ce3b8830")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 4, dtype family float, shape (2, 3, 33, 33) 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)

# Create a small image: 3 color channels, height 64, width 48
image = np.random.rand(64, 48, 3).astype(np.float32)

# Create a few patch coordinate positions (i, j)
# These should be within the image bounds
ij = [[10, 15], [30, 20]]

device = 'cpu'

ARGS = {
    'image': image,
    'ij': ij,
    'device': device,
    'patch_size': 33
}
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 4, dtype family float, shape (2, 3, 33, 33) — shape (2, 3, 33, 33) 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
generate_empty_reconstruction Ran
ran on a synthesized input
mihaidusmanu/local-feature-refinement/reconstruction-scripts/colmap_utils.py
code served (permissive licence) · get_code("e82ecebaf174cd8b")
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 os
import tempfile
import shutil

# Create temporary directories for testing
temp_dir = tempfile.mkdtemp()
reference_model_path = os.path.join(temp_dir, 'reference_model')
empty_model_path = os.path.join(temp_dir, 'empty_model')

os.makedirs(reference_model_path, exist_ok=True)

# Create cameras.txt with header and sample data
cameras_txt = os.path.join(reference_model_path, 'cameras.txt')
with open(cameras_txt, 'w') as f:
    f.write('# Header line 1\n')
    f.write('# Header line 2\n')
    f.write('# Header line 3\n')
    f.write('1 PINHOLE 640 480 500.0 320.0 240.0\n')
    f.write('2 PINHOLE 640 480 500.0 320.0 240.0\n')

# Create images.txt with header and sample data
images_txt = os.path.join(reference_model_path, 'images.txt')
with open(images_txt, 'w') as f:
    f.write('# Header line 1\n')
    f.write('# Header line 2\n')
    f.write('# Header line 3\n')
    f.write('# Header line 4\n')
    f.write('1 0.0 0.0 0.0 0.0 0.0 0.0 1.0 1 image1.jpg\n')
    f.write('\n')
    f.write('2 0.1 0.1 0.1 0.1 0.1 0.1 1.0 2 image2.jpg\n')
    f.write('\n')

ARGS = {
    'reference_model_path': reference_model_path,
    'empty_model_path': empty_model_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 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
grid_positions Ran
ran on a synthesized input
mihaidusmanu/local-feature-refinement/two-view-refinement/refinement.py
code served (permissive licence) · get_code("edd3e2e05c9973fe")
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 (4, 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)

ARGS = {
    "shape": (4, 3),
    "device": torch.device("cpu")
}
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 (4, 3, 2) — shape (4, 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
image_ids_to_pair_id Ran
ran on a synthesized input
mihaidusmanu/local-feature-refinement/reconstruction-scripts/colmap_utils.py
code served (permissive licence) · get_code("e25eda10f267f95a")
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 int, ndim 0, dtype family int, 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)

# Analyze the function:
# - Takes two parameters: image_id1 and image_id2
# - Both are used in numeric comparisons (>) and arithmetic operations (* and +)
# - They should be integers or numeric types

ARGS = {
    "image_id1": 5,
    "image_id2": 3,
}
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 int, ndim 0, dtype family int, 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
mnn_ratio_matcher Ran
ran on a synthesized input
mihaidusmanu/local-feature-refinement/two-view-refinement/feature_matchers.py
code served (permissive licence) · get_code("05462c92980d7aec")
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 torch

torch.manual_seed(0)

# Create small descriptor tensors
# descriptors1: shape (2, 4) - 2 descriptors of dimension 4
descriptors1 = torch.randn(2, 4)

# descriptors2: shape (3, 4) - 3 descriptors of dimension 4
descriptors2 = torch.randn(3, 4)

# ratio parameter for Lowe's ratio test
ratio = 0.8

ARGS = {
    "descriptors1": descriptors1,
    "descriptors2": descriptors2,
    "ratio": ratio,
}
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 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
mnn_similarity_matcher Ran
ran on a synthesized input
mihaidusmanu/local-feature-refinement/two-view-refinement/feature_matchers.py
code served (permissive licence) · get_code("3d3ffa36eb5a0e0c")
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 torch

torch.manual_seed(0)

# Create two small descriptor tensors with distinct shapes for clarity
# descriptors1: 2 descriptors of dimension 4
# descriptors2: 3 descriptors of dimension 4
descriptors1 = torch.randn(2, 4)
descriptors2 = torch.randn(3, 4)

# Define threshold parameter
threshold = 0.8

ARGS = {
    "descriptors1": descriptors1,
    "descriptors2": descriptors2,
    "threshold": threshold,
}
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 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
recover_images Ran
ran on a synthesized input
mihaidusmanu/local-feature-refinement/local-feature-evaluation/compare_reconstructions.py
code served (permissive licence) · get_code("c036dbf1a0125eb0")
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 os
import tempfile

# Create a temporary directory structure with the expected file format
temp_dir = tempfile.mkdtemp()
scene_path = temp_dir

# Create images.txt with the expected format
# First 4 lines are headers, then alternating image data lines and blank lines
images_txt_content = """# Image list with two lines of data per image:
# IMAGE_ID, QW, QX, QY, QZ, TX, TY, TZ, CAMERA_ID, NAME
# POINTS2D[] as (x, y, point3D_id)
# Number of images: 2
1 0.5 0.0 0.0 0.5 1.0 2.0 3.0 1 image1.jpg
0.1 0.2 1 0.3 0.4 2 0.5 0.6 3
2 0.6 0.1 0.1 0.4 1.5 2.5 3.5 1 image2.jpg
0.2 0.3 1 0.4 0.5 2 0.6 0.7 3
"""

images_file_path = os.path.join(scene_path, 'images.txt')
with open(images_file_path, 'w') as f:
    f.write(images_txt_content)

ARGS = {
    'scene_path': scene_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 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
array_to_blob Not yet run
unverified (harvested, no recorded run)
mihaidusmanu/local-feature-refinement/reconstruction-scripts/colmap_utils.py
code served (permissive licence) · get_code("2ba4e2304dc3100f")
How it ran

no invocation recorded.

estimate_displacements Not yet run
unverified (harvested, no recorded run)
mihaidusmanu/local-feature-refinement/two-view-refinement/refinement.py
code served (permissive licence) · get_code("b6dd46fbe440a533")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 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 9 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 2026-11-21.

Repositories linked to this paper

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

Abstract

In this work, we address the problem of refining the geometry of local image features from multiple views without known scene or camera geometry. Current approaches to local feature detection are inherently limited in their keypoint localization accuracy because they only operate on a single view. This limitation has a negative impact on downstream tasks such as Structure-from-Motion, where inaccurate keypoints lead to large errors in triangulation and camera localization. Our proposed method naturally complements the traditional feature extraction and matching paradigm. We first estimate local geometric transformations between tentative matches and then optimize the keypoint locations over multiple views jointly according to a non-linear least squares formulation. Throughout a variety of experiments, we show that our method consistently improves the triangulation and camera localization performance for both hand-crafted and learned local features.

For agents

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

get_harvested_code_for_paper("2003.08348")
get_code_for_paper("2003.08348")
have("2003.08348")

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

curl https://syntology.ai/api/ran/2003.08348.json

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

[![Syntology run record](https://syntology.ai/api/ran/2003.08348.svg)](https://syntology.ai/paper/2003.08348)

Connect an agent — have() is free.