SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.04250 · 2020

AdaLAM: Revisiting Handcrafted Outlier Detection

arXiv · PDF · Open in the Atlas

Code that ran

13 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 13 have no contract verdict; 0 failures of our instrument; 6 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/2006.04250.json · badge: /api/ran/2006.04250.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
gujingxiao/BaiduStar2020-Traffic-Sign-Detection-And-Pair-Competition-Solution community (classifier label)
classifier basis not recorded on the link
6 ran · 0 no recorded run
antonbrazhnyi/AdaLAM community (classifier label)
classifier basis not recorded on the link
4 ran · 5 no recorded run
cavalli1234/AdaLAM community (classifier label)
classifier basis not recorded on the link
3 ran · 1 no recorded run
FunctionStatusWhere it lives
arange_sequence Ran
ran on a synthesized input
antonbrazhnyi/AdaLAM/adalam/utils.py
code served (permissive licence) · get_code("273031afa673ed50")
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 1, dtype family int, shape (10,) 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 takes a 'ranges' parameter which is a tensor
# Looking at the function body:
# - ranges.device is accessed (so it's a tensor)
# - torch.max(ranges).item() is called
# - ranges.shape[0] is accessed
# - ranges.unsqueeze(-1) is called
# So ranges should be a 1D tensor of integers

ranges = torch.tensor([2, 5, 1, 2], dtype=torch.long)

ARGS = {
    "ranges": ranges
}
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 1, dtype family int, shape (10,) — shape (10,) 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
dist_matrix Ran
ran on a synthesized input
antonbrazhnyi/AdaLAM/adalam/utils.py
code served (permissive licence) · get_code("a3a952583dcee63e")
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, 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 torch.nn as nn

torch.manual_seed(0)

# The function expects:
# - d1: a tensor that supports ** (power), .sum(), .unsqueeze(), .transpose(), and @ (matmul)
# - d2: a tensor that supports .transpose() and @ (matmul)
# - is_normalized: a boolean

# Create small tensors with distinct sizes for clarity
# d1 and d2 should have shape (batch, ..., dim) based on the operations
# .transpose(1, 2) suggests 3D tensors: (batch, seq_len, dim)

batch_size = 2
seq_len_d1 = 3
seq_len_d2 = 4
dim = 5

d1 = torch.randn(batch_size, seq_len_d1, dim)
d2 = torch.randn(batch_size, seq_len_d2, dim)

is_normalized = False

ARGS = {
    "d1": d1,
    "d2": d2,
    "is_normalized": is_normalized
}
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, 4) — shape (2, 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
dist_matrix Ran
ran on a synthesized input
cavalli1234/AdaLAM/adalam/utils.py
code served (permissive licence) · get_code("044e5f3c6e44c154")
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, 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 torch

torch.manual_seed(0)

# d1 and d2 are tensors that will be used in matrix operations
# The function computes distances between two sets of vectors
# d1 @ d2.t() performs matrix multiplication, so d1 and d2 must be 2D
# (d1 ** 2).sum(1) sums across dimension 1, so we need at least 2D tensors

d1 = torch.randn(2, 4)  # batch_size=2, dim=4
d2 = torch.randn(3, 4)  # batch_size=3, dim=4 (different size to show distinction)

is_normalized = False

ARGS = {
    "d1": d1,
    "d2": d2,
    "is_normalized": is_normalized,
}
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, 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
distort_color Ran
ran on a synthesized input
gujingxiao/BaiduStar2020-Traffic-Sign-Detection-And-Pair-Competition-Solution/metric_learning_traffic/imgtool.py
code served (permissive licence) · get_code("aa62684e7175e3a0")
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, 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 torch
import torch.nn as nn

torch.manual_seed(0)

# The function distort_color takes an img parameter and returns it unchanged
# img should be a tensor representing an image
# Create a small image tensor (batch_size=2, channels=3, height=4, width=4)
img = torch.randn(2, 3, 4, 4)

ARGS = {
    "img": img
}
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 4, dtype family float, shape (2, 3, 4, 4) — shape (2, 3, 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
extract_local_patterns Ran
ran on a synthesized input
cavalli1234/AdaLAM/adalam/core.py
code served (permissive licence) · get_code("ccf718aa4ea2a54a")
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 tensors for testing
num_keypoints_im1 = 5
num_keypoints_im2 = 5
num_seeds = 2
num_matches = 8  # Number of true correspondences in the boolean matrix

# fnn12: Maps keypoint indices from image 1 to image 2
# For 5 keypoints in im1, we need indices 0-4 pointing to im2 keypoints
fnn12 = torch.tensor([1, 2, 0, 3, 4], dtype=torch.long)

# fnn_to_seed_local_consistency_map_corr: Boolean matrix (num_seeds, num_keypoints)
# Shape (2, 5) with True values indicating which keypoints belong to which seed
fnn_to_seed_local_consistency_map_corr = torch.tensor([
    [True, True, True, True, False],
    [False, False, True, False, True]
], dtype=torch.bool)

# k1: Keypoint locations in image I_1
# Shape (num_keypoints_im1, 2) for 2D coordinates
k1 = torch.tensor([
    [10.0, 20.0],
    [15.0, 25.0],
    [30.0, 40.0],
    [50.0, 60.0],
    [100.0, 120.0]
], dtype=torch.float32)

# k2: Keypoint locations in image I_2
# Shape (num_keypoints_im2, 2) for 2D coordinates
k2 = torch.tensor([
    [11.0, 21.0],
    [16.0, 26.0],
    [31.0, 41.0],
    [51.0, 61.0],
    [101.0, 121.0]
], dtype=torch.float32)

# im1seeds: Keypoint indices of chosen seeds in image I_1
im1seeds = torch.tensor([0, 2], dtype=torch.long)

# im2seeds: Keypoint indices of chosen seeds in image I_2
im2seeds = torch.tensor([1, 0], dtype=torch.long)

# scores: Confidence scores for correspondences (lower is better)
# Shape (num_keypoints_im1,) with values between 0 and 1
scores = torch.tensor([0.1, 0.3, 0.2, 0.15, 0.25], dtype=torch.float32)

ARGS = {
    'fnn12': fnn12,
    'fnn_to_seed_local_consistency_map_corr': fnn_to_seed_local_consistency_map_corr,
    'k1': k1,
    'k2': k2,
    'im1seeds': im1seeds,
    'im2seeds': im2seeds,
    'scores': scores
}
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
generate_index Ran
ran on a synthesized input
gujingxiao/BaiduStar2020-Traffic-Sign-Detection-And-Pair-Competition-Solution/metric_learning_traffic/losses/commonfunc.py
code served (permissive licence) · get_code("8934a86d67997100")
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 (16,) 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

# Analyze the function to understand parameters:
# - batch_size: used to create array of size batch_size * batch_size, reshape to (-1, batch_size)
#   and for division operations (batch_size // samples_each_class)
# - samples_each_class: used for division with batch_size

# The function expects batch_size to be divisible by samples_each_class
# Let's use small values: batch_size=4, samples_each_class=2

ARGS = {
    "batch_size": 4,
    "samples_each_class": 2
}
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 int, shape (16,) — shape (16,) 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
orientation_diff Ran
ran on a synthesized input
antonbrazhnyi/AdaLAM/adalam/utils.py
code served (permissive licence) · get_code("6e21a7f85e0bd425")
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 1, dtype family float, 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 torch
import numpy as np

torch.manual_seed(0)

# The function takes two orientation values (o1, o2) and computes the difference
# normalized to the range [-180, 180). The function uses subtraction, comparison,
# and addition operations, which work on numpy arrays, torch tensors, or scalars.

# We'll use torch tensors as they support the required operations
o1 = torch.tensor([10.0, 45.0, 170.0, 350.0])
o2 = torch.tensor([20.0, 200.0, 190.0, 10.0])

ARGS = {
    "o1": o1,
    "o2": o2,
}
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 1, dtype family float, 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
postprocess Ran
ran on a synthesized input
gujingxiao/BaiduStar2020-Traffic-Sign-Detection-And-Pair-Competition-Solution/PaddleClas/traffic_class_infer.py
code served (permissive licence) · get_code("9e2a7fd85f0594c2")
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 numpy as np
import torch

torch.manual_seed(0)

# The function expects outputs to be indexable (outputs[0])
# and outputs[0] to be convertible to a numpy array and flattened
# It uses argsort, slicing, and indexing on the probability array

# Create a simple nested structure where outputs[0] is a list/array of probabilities
outputs = [np.array([0.1, 0.5, 0.3, 0.05, 0.05])]

ARGS = {
    "outputs": outputs,
    "topk": 2
}
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 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
preprocess Ran
ran on a synthesized input
gujingxiao/BaiduStar2020-Traffic-Sign-Detection-And-Pair-Competition-Solution/PaddleClas/traffic_class_infer.py
code served (permissive licence) · get_code("63e249d1dde56372")
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 4, dtype family float, shape (2, 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 torch
import numpy as np

torch.manual_seed(0)

# The function expects:
# - data: a 3D array-like that supports slicing with [:, :, ::-1]
# - ops: an iterable of callables that each take data and return transformed data

# Create a small 3D numpy array (batch=2, height=4, width=3, channels=3)
# Using ::-1 on last dimension reverses color channels (e.g., BGR to RGB)
data = np.random.randn(2, 4, 3, 3).astype(np.float32)

# Create simple operation callables that transform the data
class NormalizeOp:
    def __call__(self, data):
        return data / 255.0

class ScaleOp:
    def __call__(self, data):
        return data * 2.0

ops = [NormalizeOp(), ScaleOp()]

ARGS = {
    "data": data,
    "ops": ops,
}
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 ndarray, ndim 4, dtype family float, shape (2, 4, 3, 3) — shape (2, 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
random_crop Ran
ran on a synthesized input
gujingxiao/BaiduStar2020-Traffic-Sign-Detection-And-Pair-Competition-Solution/metric_learning_traffic/imgtool.py
code served (permissive licence) · get_code("9b681e8a76479f0f")
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 (64, 64, 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 math
import random
import numpy as np

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

# Create a small test image (height=100, width=80, channels=3)
# Must be a numpy array with shape (height, width, channels)
img = np.random.randint(0, 256, size=(100, 80, 3), dtype=np.uint8)

# Define arguments for random_crop function
ARGS = {
    'img': img,
    'size': 64,
    'scale': [0.08, 1.0],
    'ratio': [3.0 / 4.0, 4.0 / 3.0]
}
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 (64, 64, 3) — shape (64, 64, 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
rotate_image Ran
ran on a synthesized input
gujingxiao/BaiduStar2020-Traffic-Sign-Detection-And-Pair-Competition-Solution/metric_learning_traffic/imgtool.py
code served (permissive licence) · get_code("3668fcd0a8a7ae06")
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
import random

random.seed(0)
np.random.seed(0)

# Create a small test image (4x3 RGB image)
img = np.random.randint(0, 256, (4, 3, 3), dtype=np.uint8)

ARGS = {
    "img": img
}
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
select_seeds Ran
ran on a synthesized input
antonbrazhnyi/AdaLAM/adalam/core.py
code served (permissive licence) · get_code("ce984335a94fc19e")
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 int; 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)

# dist1: distance matrix between keypoints in image I_1
# Shape: (batch_size, num_keypoints_in_image1, num_keypoints_in_image1)
batch_size = 2
num_keypoints = 4
dist1 = torch.rand(batch_size, num_keypoints, num_keypoints)

# R1: base radius of neighborhoods in image I_1
# Shape: (batch_size,)
R1 = torch.tensor([0.5, 0.6])

# scores1: confidence scores on putative matches (Lowe's ratio scores)
# Shape: (batch_size, num_keypoints_in_image1)
scores1 = torch.rand(batch_size, num_keypoints)

# fnn12: matches between keypoints of I_1 and I_2
# The i-th entry is j if keypoint k_i in I_1 is matched to keypoint k_j in I_2
# Shape: (batch_size, num_keypoints_in_image1)
# Values should be valid indices into image 2 keypoints
num_keypoints_im2 = 5
fnn12 = torch.randint(0, num_keypoints_im2, (batch_size, num_keypoints))

# mnn: mask indicating which putative matches are mutual nearest neighbors
# Shape: (batch_size, num_keypoints_in_image1)
mnn = torch.randint(0, 2, (batch_size, num_keypoints)).bool()

ARGS = {
    "dist1": dist1,
    "R1": R1,
    "scores1": scores1,
    "fnn12": fnn12,
    "mnn": mnn,
}
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 int — 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
select_seeds Ran
ran on a synthesized input
cavalli1234/AdaLAM/adalam/core.py
code served (permissive licence) · get_code("ca2ddc97cec6ba7b")
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 int; 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 the function inputs
num_keypoints_1 = 5
num_keypoints_2 = 6

# dist1: Precomputed distance matrix between keypoints in image I_1
# Shape: (num_keypoints_1, num_keypoints_1)
dist1 = torch.tensor([
    [0.0, 0.5, 2.0, 3.5, 1.2],
    [0.5, 0.0, 1.8, 2.0, 0.8],
    [2.0, 1.8, 0.0, 1.5, 2.5],
    [3.5, 2.0, 1.5, 0.0, 3.0],
    [1.2, 0.8, 2.5, 3.0, 0.0]
], dtype=torch.float32)

# R1: Base radius of neighborhoods in image I_1
R1 = 1.5

# scores1: Confidence scores on putative matches (Lowe's ratio scores)
# Shape: (num_keypoints_1,)
scores1 = torch.tensor([0.5, 0.3, 0.7, 0.2, 0.6], dtype=torch.float32)

# fnn12: Matches between keypoints of I_1 and I_2
# The i-th entry is j if keypoint k_i in I_1 is matched to keypoint k_j in I_2
# Shape: (num_keypoints_1,)
fnn12 = torch.tensor([1, 2, 3, 4, 5], dtype=torch.long)

# mnn: Mask indicating which putative matches are mutual nearest neighbors
# Shape: (num_keypoints_1,)
mnn = torch.tensor([True, False, True, False, True], dtype=torch.bool)

ARGS = {
    "dist1": dist1,
    "R1": R1,
    "scores1": scores1,
    "fnn12": fnn12,
    "mnn": mnn
}
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 int — 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
confidence_based_inlier_selection Not yet run
unverified (harvested, no recorded run)
antonbrazhnyi/AdaLAM/adalam/ransac.py
code served (permissive licence) · get_code("0b54a05cf65bbe27")
How it ran

no invocation recorded.

extract_local_patterns Not yet run
unverified (harvested, no recorded run)
antonbrazhnyi/AdaLAM/adalam/core.py
code served (permissive licence) · get_code("1baf1bbcafd8baa2")
How it ran

no invocation recorded.

extract_neighborhood_sets Not yet run
unverified (harvested, no recorded run)
antonbrazhnyi/AdaLAM/adalam/core.py
code served (permissive licence) · get_code("4c3eec786c61959e")
How it ran

no invocation recorded.

extract_neighborhood_sets Not yet run
unverified (harvested, no recorded run)
cavalli1234/AdaLAM/adalam/core.py
code served (permissive licence) · get_code("9e1bd9727a566275")
How it ran

no invocation recorded.

group_sum_and_cumsum Not yet run
unverified (harvested, no recorded run)
antonbrazhnyi/AdaLAM/adalam/ransac.py
code served (permissive licence) · get_code("6635220d02d11dda")
How it ran

no invocation recorded.

stable_sort_residuals Not yet run
unverified (harvested, no recorded run)
antonbrazhnyi/AdaLAM/adalam/ransac.py
code served (permissive licence) · get_code("6af9bdca2053ef9c")
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

Local feature matching is a critical component of many computer vision pipelines, including among others Structure-from-Motion, SLAM, and Visual Localization. However, due to limitations in the descriptors, raw matches are often contaminated by a majority of outliers. As a result, outlier detection is a fundamental problem in computer vision, and a wide range of approaches have been proposed over the last decades. In this paper we revisit handcrafted approaches to outlier filtering. Based on best practices, we propose a hierarchical pipeline for effective outlier detection as well as integrate novel ideas which in sum lead to AdaLAM, an efficient and competitive approach to outlier rejection. AdaLAM is designed to effectively exploit modern parallel hardware, resulting in a very fast, yet very accurate, outlier filter. We validate AdaLAM on multiple large and diverse datasets, and we submit to the Image Matching Challenge (CVPR2020), obtaining competitive results with simple baseline descriptors. We show that AdaLAM is more than competitive to current state of the art, both in terms of efficiency and effectiveness.

For agents

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

get_harvested_code_for_paper("2006.04250")
get_code_for_paper("2006.04250")
have("2006.04250")

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

curl https://syntology.ai/api/ran/2006.04250.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.04250.svg)](https://syntology.ai/paper/2006.04250)

Connect an agent — have() is free.