SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1910.02190 · 2019

Kornia: an Open Source Differentiable Computer Vision Library for PyTorch

arXiv · PDF · Open in the Atlas

Code that ran

12 of 20 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 12 have no contract verdict; 0 failures of our instrument; 8 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/1910.02190.json · badge: /api/ran/1910.02190.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
arraiyopensource/torchgeometry community (classifier label)
classifier basis not recorded on the link
6 ran · 8 no recorded run
manyids2/kornia community (classifier label)
classifier basis not recorded on the link
6 ran · 0 no recorded run
FunctionStatusWhere it lives
KORNIA_UNWRAP Ran
ran on a synthesized input
arraiyopensource/torchgeometry/kornia/core/check.py
code served (permissive licence) · get_code("599c18b8f8db01de")
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
from typing import Any

torch.manual_seed(0)

# The function KORNIA_UNWRAP is a generic unwrapping utility that takes
# an object and a type, and returns it cast to that type.
# It's a simple pass-through with type casting.
# We need to provide:
# 1. maybe_obj: Any object to unwrap
# 2. typ: The expected type after unwrap

# Create a simple object to unwrap
maybe_obj = torch.tensor([1.0, 2.0, 3.0, 4.0])

# Define the expected type
typ = torch.Tensor

ARGS = {
    "maybe_obj": maybe_obj,
    "typ": typ,
}
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 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
add_weighted Ran
ran on a synthesized input
manyids2/kornia/kornia/enhance/core.py
code served (permissive licence) · get_code("98daa2eaacc24301")
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, 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

torch.manual_seed(0)

# Create two small tensors for blending
src1 = torch.randn(2, 4)
src2 = torch.randn(2, 4)

# Define scalar weights and bias
alpha = 0.7
beta = 0.3
gamma = 0.1

ARGS = {
    'src1': src1,
    'alpha': alpha,
    'src2': src2,
    'beta': beta,
    'gamma': gamma,
}
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, 4) — shape (2, 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
hf_url Ran
ran on a synthesized input
arraiyopensource/torchgeometry/kornia/core/download.py
code served (permissive licence) · get_code("eee8227d7d06c693")
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 str. 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

# Call hf_url with different repository and filename combinations
ARGS = {
    'repo': 'hardnet',
    'filename': 'HardNetPP.pth'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
inverse_depth_smoothness_loss Ran
ran on a synthesized input
manyids2/kornia/kornia/losses/depth_smooth.py
code served (permissive licence) · get_code("451611489a552dad")
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 inverse depth tensor: BxCxHxW
# B=2 (batch), C=1 (channels), H=4 (height), W=3 (width)
idepth = torch.randn(2, 1, 4, 3, dtype=torch.float32)

# Create image tensor with same spatial dimensions but different channel count
# B=2 (batch), C=3 (RGB channels), H=4 (height), W=3 (width)
image = torch.randn(2, 3, 4, 3, dtype=torch.float32)

ARGS = {
    'idepth': idepth,
    'image': image,
}
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
js_div_loss_2d Ran
ran on a synthesized input
manyids2/kornia/kornia/losses/divergence.py
code served (permissive licence) · get_code("9f91916eaedf7642")
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 tiny 4D tensors: (batch=2, channels=4, height=3, width=5)
# These represent heatmaps that should sum to 1 along spatial dimensions for divergence
batch_size = 2
num_channels = 4
height = 3
width = 5

# Create input tensor - probability-like values
input_tensor = torch.rand(batch_size, num_channels, height, width)
input_tensor = input_tensor / input_tensor.sum(dim=(2, 3), keepdim=True)  # normalize

# Create target tensor - probability-like values
target_tensor = torch.rand(batch_size, num_channels, height, width)
target_tensor = target_tensor / target_tensor.sum(dim=(2, 3), keepdim=True)  # normalize

ARGS = {
    'input': input_tensor,
    'target': target_tensor,
    'reduction': 'mean'
}
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
kl_div_loss_2d Ran
ran on a synthesized input
manyids2/kornia/kornia/losses/divergence.py
code served (permissive licence) · get_code("53589e722f69b7dd")
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 input and target tensors with shape (B, N, H, W)
# Using distinct sizes: batch=2, channels=4, height=3, width=5
batch = 2
chans = 4
height = 3
width = 5

# Create valid probability distributions (non-negative, sum to 1 along spatial dims)
input_tensor = torch.rand(batch, chans, height, width)
input_tensor = input_tensor / input_tensor.sum(dim=(2, 3), keepdim=True)

target_tensor = torch.rand(batch, chans, height, width)
target_tensor = target_tensor / target_tensor.sum(dim=(2, 3), keepdim=True)

ARGS = {
    'input': input_tensor,
    'target': target_tensor,
    'reduction': 'mean'
}
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
psnr_loss Ran
ran on a synthesized input
manyids2/kornia/kornia/losses/psnr.py
code served (permissive licence) · get_code("5770f772f4f42d7f")
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
from torch.nn.functional import mse_loss

torch.manual_seed(0)

# Create small distinct tensors for input and target
input_tensor = torch.randn(2, 4)
target_tensor = torch.randn(2, 4)

# max_val is a float representing the maximum possible pixel value (e.g., 255 for 8-bit images, 1.0 for normalized images)
max_val = 1.0

ARGS = {
    'input': input_tensor,
    'target': target_tensor,
    'max_val': max_val,
}
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
time_us Ran
ran on a synthesized input
arraiyopensource/torchgeometry/benchmarks/common.py
code served (permissive licence) · get_code("28254b28b6b25aa8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import torch.utils.benchmark as bench

torch.manual_seed(0)

# Simple callable that performs a lightweight operation
def simple_fn():
    x = torch.randn(2, 4)
    y = torch.randn(2, 4)
    return (x + y).sum()

# Optional sync function for device synchronization
def sync_fn():
    if torch.cuda.is_available():
        torch.cuda.synchronize()

ARGS = {
    'fn': simple_fn,
    'min_run_time': 0.1,
    'sync': None,
    'num_threads': 1
}
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 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
time_us_or_error Ran
ran on a synthesized input
arraiyopensource/torchgeometry/benchmarks/common.py
code served (permissive licence) · get_code("ffad1746a07855b2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import torch.utils.benchmark as bench

torch.manual_seed(0)

# Define a simple function to time
def simple_fn():
    x = torch.randn(2, 4)
    y = torch.randn(2, 4)
    z = x + y
    return z

# Define an optional sync function
def sync_fn():
    torch.cuda.synchronize() if torch.cuda.is_available() else None

ARGS = {
    'fn': simple_fn,
    'min_run_time': 0.1,
    'sync': None,
    'num_threads': 2,
}
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 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
total_variation Ran
ran on a synthesized input
manyids2/kornia/kornia/losses/total_variation.py
code served (permissive licence) · get_code("ace65b30a9175199")
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 (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)

# Construct a valid image tensor for total_variation
# The function expects either 3D or 4D tensors
# 3D: (C, H, W) - single image with channels, height, width
# 4D: (B, C, H, W) - batch of images
# Using 4D tensor: batch_size=2, channels=3, height=4, width=5
img = torch.randn(2, 3, 4, 5)

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 1, dtype family float, shape (2,) — shape (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
validate_artefact Ran
ran on a synthesized input
arraiyopensource/torchgeometry/benchmarks/results_schema.py
code served (permissive licence) · get_code("3286740a3fa2ea7e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import tempfile
from pathlib import Path

# Create a temporary directory for test files
temp_dir = tempfile.TemporaryDirectory()
temp_path = Path(temp_dir.name)

# Create a valid artefact file
valid_payload = {
    "metadata": {
        "timestamp_utc": "2024-01-01T00:00:00Z",
        "git_commit": "abc123def456",
        "platform": "linux",
        "python": "3.10.0",
        "torch": "2.0.0",
        "kornia": "0.7.0",
        "device": "cpu",
        "load": {"metric1": 1.5, "metric2": None}
    },
    "results": [
        {
            "op": "conv2d",
            "backend": "cpu",
            "batch": 32,
            "median_us": 100.5,
            "throughput_per_s": 1000
        },
        {
            "op": "matmul",
            "backend": "cpu",
            "batch": 16,
            "median_us": 50.0,
            "throughput_per_s": 2000
        }
    ]
}

# Write the valid artefact to a file
artefact_path = temp_path / "test_artefact.json"
artefact_path.write_text(json.dumps(valid_payload))

ARGS = {
    "path": artefact_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
validate_result Ran
ran on a synthesized input
arraiyopensource/torchgeometry/benchmarks/results_schema.py
code served (permissive licence) · get_code("2beca07bad1498c3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import tempfile
from pathlib import Path

# Create a temporary directory and a valid result file
temp_dir = tempfile.mkdtemp()
temp_path = Path(temp_dir)

# Create a valid result JSON file with the required structure
valid_payload = {
    "metadata": {
        "timestamp_utc": "2024-01-01T00:00:00Z",
        "git_commit": "abc123def456",
        "platform": "Linux",
        "python": "3.10.0",
        "torch": "2.0.0",
        "kornia": "0.7.0",
        "device": "cpu:0",
        "load": {"model_size_mb": 100.5, "batch_time_ms": 50}
    },
    "results": [
        {
            "op": "convolution",
            "backend": "pytorch",
            "batch": 2,
            "median_us": 1500.75,
            "throughput_per_s": 1333.33
        },
        {
            "op": "matmul",
            "backend": "pytorch",
            "batch": 4,
            "median_us": 2000.5,
            "throughput_per_s": 2000.0
        }
    ]
}

# Write the payload to a file with the correct naming format
# Format: <suite>--<machine-slug>--<device>.json
result_file = temp_path / "cpu--test-machine--cpu.json"
result_file.write_text(json.dumps(valid_payload))

# Create the version subdirectory to match the kornia metadata
version_dir = temp_path / "0.7.0"
version_dir.mkdir(exist_ok=True)
versioned_result_file = version_dir / "cpu--test-machine--cpu.json"
versioned_result_file.write_text(json.dumps(valid_payload))

ARGS = {
    "path": versioned_result_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
KORNIA_CHECK Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/kornia/core/check.py
code served (permissive licence) · get_code("e8f7a3ea874a438a")
How it ran

no invocation recorded.

KORNIA_CHECK_SHAPE Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/kornia/core/check.py
code served (permissive licence) · get_code("bbf98733bfa5c36e")
How it ran

no invocation recorded.

download_file_from_url Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/kornia/core/download.py
code served (permissive licence) · get_code("62980aa194af6a33")
How it ran

no invocation recorded.

load_state_dict_from_url Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/kornia/core/download.py
code served (permissive licence) · get_code("0dfafa22115f9ffe")
How it ran

no invocation recorded.

run_metadata Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/benchmarks/common.py
code served (permissive licence) · get_code("3d19faa7fd35b5b9")
How it ran

no invocation recorded.

torch_version_ge Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/kornia/core/_compat.py
code served (permissive licence) · get_code("c962bead721155d5")
How it ran

no invocation recorded.

torch_version_le Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/kornia/core/_compat.py
code served (permissive licence) · get_code("d55371b833e21170")
How it ran

no invocation recorded.

torch_version_lt Not yet run
unverified (harvested, no recorded run)
arraiyopensource/torchgeometry/kornia/core/_compat.py
code served (permissive licence) · get_code("44c8e0b0e11a0d7e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 20 samples lifted from this paper's linked repositories, 20 carry a function identity and 0 do not. Of the 20 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 20 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

This work presents Kornia -- an open source computer vision library which consists of a set of differentiable routines and modules to solve generic computer vision problems. The package uses PyTorch as its main backend both for efficiency and to take advantage of the reverse-mode auto-differentiation to define and compute the gradient of complex functions. Inspired by OpenCV, Kornia is composed of a set of modules containing operators that can be inserted inside neural networks to train models to perform image transformations, camera calibration, epipolar geometry, and low level image processing techniques, such as filtering and edge detection that operate directly on high dimensional tensor representations. Examples of classical vision problems implemented using our framework are provided including a benchmark comparing to existing vision libraries.

For agents

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

get_harvested_code_for_paper("1910.02190")
get_code_for_paper("1910.02190")
have("1910.02190")

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

curl https://syntology.ai/api/ran/1910.02190.json

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

[![Syntology run record](https://syntology.ai/api/ran/1910.02190.svg)](https://syntology.ai/paper/1910.02190)

Connect an agent — have() is free.