SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1908.05547 · 2019

Beyond Cartesian Representations for Local Descriptors

arXiv · PDF · Open in the Atlas

Code that ran

9 of 11 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 8 have no contract verdict; 1 failure 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/1908.05547.json · badge: /api/ran/1908.05547.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
cvlab-epfl/log-polar-descriptors canonical (classifier label)
classifier basis: census
9 ran · 2 no recorded run
FunctionStatusWhere it lives
ErrorRateAt95Recall Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/modules/hardnet/eval_metrics.py
code served (permissive licence) · get_code("d5f2c58e85e22857")
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 float, 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 numpy as np

# The function takes labels and scores as numpy arrays
# labels: binary labels (0 or 1) indicating negative/positive examples
# scores: confidence scores for each example

# Create small test data
np.random.seed(0)

# Generate binary labels (0s and 1s)
labels = np.array([1, 0, 1, 1, 0, 1, 0, 1], dtype=np.int32)

# Generate scores between 0 and 1
scores = np.array([0.9, 0.2, 0.8, 0.85, 0.1, 0.95, 0.3, 0.7], dtype=np.float32)

ARGS = {
    "labels": labels,
    "scores": scores
}
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 float, 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
distance_matrix_vector Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/modules/hardnet/losses.py
code served (permissive licence) · get_code("af942949113ac97f")
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, 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)

# The function expects:
# - anchor: a 2D tensor of shape (batch_size, descriptor_dim)
# - positive: a 2D tensor of shape (batch_size, descriptor_dim)
# Both are used with:
#   - torch.sum(..., dim=1) - summing across dimension 1
#   - .unsqueeze(-1) - adding a dimension
#   - .size(0) - getting batch size
#   - .repeat() - repeating values
#   - torch.t() - transposing
#   - torch.bmm() - batch matrix multiply
# So we need 2D tensors with matching descriptor dimensions

anchor = torch.randn(2, 4)  # batch_size=2, descriptor_dim=4
positive = torch.randn(2, 4)  # batch_size=2, descriptor_dim=4

ARGS = {
    "anchor": anchor,
    "positive": positive,
}
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, 2) — shape (2, 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
distance_vectors_pairwise Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/modules/hardnet/losses.py
code served (permissive licence) · get_code("ad7f74d1717ab5c5")
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 torch

torch.manual_seed(0)

# Create small tensors for anchor, positive, and negative
# Each tensor should have shape (batch_size, feature_dim)
# Using batch_size=2 and feature_dim=4

anchor = torch.randn(2, 4)
positive = torch.randn(2, 4)
negative = torch.randn(2, 4)

ARGS = {
    "anchor": anchor,
    "positive": positive,
    "negative": negative,
}
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, 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
extract_deep_features_directly Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/baselines/geodesc/image_matching_utils.py
code served (permissive licence) · get_code("493ee778ced3888d")
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 (1536, 8) 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 threading import Thread
from queue import Queue

class MockSession:
    """Mock TensorFlow session that simulates the required behavior."""
    def run(self, output_name, feed_dict=None):
        # Get the input patch data
        input_data = feed_dict.get('input:0')
        # Return some dummy feature data with shape (batch_size, feature_dim)
        batch_size = input_data.shape[0]
        return np.random.randn(batch_size, 8).astype(np.float32)

# Create patches: shape (1536, 4) - 1536 patches of 4-dimensional data
# This gives us 3 full batches (512 each) plus a remainder
patches = np.random.randn(1536, 4).astype(np.float32)

# Create a mock session
sess = MockSession()

# Set qtz parameter (quantize flag)
qtz = False

ARGS = {
    'sess': sess,
    'patches': patches,
    'qtz': qtz
}
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 (1536, 8) — shape (1536, 8) 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
loss_random_sampling Ran
ran; our contract draft was wrong, not the code
cvlab-epfl/log-polar-descriptors/modules/hardnet/losses.py
code served (permissive licence) · get_code("d20d4ac66c6e73d6")
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)

# Create small tensors for anchor, positive, and negative
# Using batch_size=2, embedding_dim=4 for clarity
batch_size = 2
embedding_dim = 4

anchor = torch.randn(batch_size, embedding_dim)
positive = torch.randn(batch_size, embedding_dim)
negative = torch.randn(batch_size, embedding_dim)

ARGS = {
    'anchor': anchor,
    'positive': positive,
    'negative': negative,
    'anchor_swap': False,
    'margin': 1.0,
    'loss_type': 'triplet_margin'
}
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
main Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/baselines/geodesc/image_matching_utils.py
code served (permissive licence) · get_code("efe5a9906bb5c441")
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 (1100, 8) 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 threading import Thread
from queue import Queue


class MockSession:
    """Mock TensorFlow session that simulates sess.run()"""
    def run(self, output_name, feed_dict=None):
        # Extract the input tensor from feed_dict
        input_data = feed_dict.get('input:0', None)
        if input_data is None:
            raise ValueError("input:0 not found in feed_dict")
        # Return a mock feature output with shape (batch_size, feature_dim)
        batch_size = input_data.shape[0]
        # Return random features with shape (batch_size, 8)
        return np.random.randn(batch_size, 8).astype(np.float32)


# Create mock session
session = MockSession()

# Create patches array - total of 1100 patches with spatial dims 4x3
# This will test the loop logic since 1100 % 512 > 0
patches = np.random.randn(1100, 4, 3).astype(np.float32)

# Define arguments dict
ARGS = {
    'argv': None,
    'session': session,
    'patches': patches
}
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 (1100, 8) — shape (1100, 8) 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
rotationMatrixToEulerAngles Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/utils/utils.py
code served (permissive licence) · get_code("7d1754d2b2a5f57d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (3,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a valid rotation matrix (3x3 orthogonal matrix with det=1)
# Using a rotation around z-axis by 45 degrees as an example
angle = np.pi / 4
cos_a = np.cos(angle)
sin_a = np.sin(angle)

R = np.array([
    [cos_a, -sin_a, 0],
    [sin_a, cos_a, 0],
    [0, 0, 1]
], dtype=np.float64)

ARGS = {
    "R": R
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (3,) — shape (3,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
splitTrainValTest Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/utils/utils.py
code served (permissive licence) · get_code("668ad7b5ada2426e")
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. 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 math
import torch

# Analyze the function:
# - data: needs to support len() and slicing [leftIDX:rightIDX]
# - splitFracts: needs to be iterable, each element used in fract * len(data)
# 
# The function splits data into chunks based on fractions.
# We'll use a simple list as data and a list of floats for splitFracts.

data = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]  # Simple list of 10 items
splitFracts = [0.6, 0.2, 0.2]  # 60% train, 20% val, 20% test

ARGS = {
    "data": data,
    "splitFracts": splitFracts,
}
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 list — a list 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
str2bool Ran
ran on a synthesized input
cvlab-epfl/log-polar-descriptors/modules/hardnet/utils.py
code served (permissive licence) · get_code("d1b2a3e236cfd907")
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 bool, 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

ARGS = {
    'v': 'true'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, 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
batch_feed Not yet run
unverified (harvested, no recorded run)
cvlab-epfl/log-polar-descriptors/utils/utils.py
code served (permissive licence) · get_code("0bad1b3406498856")
How it ran

no invocation recorded.

show_images Not yet run
unverified (harvested, no recorded run)
cvlab-epfl/log-polar-descriptors/modules/hardnet/utils.py
code served (permissive licence) · get_code("a32b0764cba8d52c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 2 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 2 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.

FunctionAlso present inEarliest-datedOthers named
loss_random_sampling 1 other paper
1 later
2006.05077 (2020-06)
SEKD: Self-Evolving Keypoint Detection and Description
str2bool 1 other paper
1 later
2203.14952 (2022-03)
Energy-based Latent Aligner for Incremental Learning

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

The dominant approach for learning local patch descriptors relies on small image regions whose scale must be properly estimated a priori by a keypoint detector. In other words, if two patches are not in correspondence, their descriptors will not match. A strategy often used to alleviate this problem is to "pool" the pixel-wise features over log-polar regions, rather than regularly spaced ones. By contrast, we propose to extract the "support region" directly with a log-polar sampling scheme. We show that this provides us with a better representation by simultaneously oversampling the immediate neighbourhood of the point and undersampling regions far away from it. We demonstrate that this representation is particularly amenable to learning descriptors with deep networks. Our models can match descriptors across a much wider range of scales than was possible before, and also leverage much larger support regions without suffering from occlusions. We report state-of-the-art results on three different datasets.

For agents

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

get_harvested_code_for_paper("1908.05547")
get_code_for_paper("1908.05547")
have("1908.05547")

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

curl https://syntology.ai/api/ran/1908.05547.json

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

[![Syntology run record](https://syntology.ai/api/ran/1908.05547.svg)](https://syntology.ai/paper/1908.05547)

Connect an agent — have() is free.