SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.09275 · ECCV · 2022

ART-SS: An Adaptive Rejection Technique for Semi-Supervised restoration for adverse weather-affected images

Carey Priebe, Vishal Patel, Rajeev Yasarla

arXiv · PDF · Open in the Atlas

Code that ran

6 of 7 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 4 failures of our instrument; 1 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/2203.09275.json · badge: /api/ran/2203.09275.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
rajeevyasarla/ART-SS community
Papers with Code archive
6 ran · 1 no recorded run
FunctionStatusWhere it lives
kernel_distance Ran
ran; our contract draft was wrong, not the code
rajeevyasarla/ART-SS/ART_SS.py
code served (permissive licence) · get_code("554c250767aa6b83")
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) 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
import torch.nn as nn

torch.manual_seed(0)

# Test case 1: 2D tensors
X_u_2d = torch.randn(2, 4)
X_l_2d = torch.randn(3, 4)

# Test case 2: 3D tensors
X_u_3d = torch.randn(2, 5, 4)
X_l_3d = torch.randn(2, 6, 4)

# Test case 3: 4D tensors
X_u_4d = torch.randn(2, 3, 5, 4)
X_l_4d = torch.randn(2, 3, 6, 4)

# Use 2D case as the primary test
ARGS = {
    "X_u": X_u_2d,
    "X_l": X_l_2d,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
kernel_linear Ran
ran; our contract draft was wrong, not the code
rajeevyasarla/ART-SS/ART_SS.py
code served (permissive licence) · get_code("0bf53920a5899333")
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) 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)

# Test case 1: 2D tensors
X_u_2d = torch.randn(2, 4)
X_l_2d = torch.randn(3, 4)

# Test case 2: 3D tensors
X_u_3d = torch.randn(2, 3, 4)
X_l_3d = torch.randn(2, 5, 4)

# Test case 3: 4D tensors
X_u_4d = torch.randn(2, 3, 4, 5)
X_l_4d = torch.randn(2, 3, 6, 5)

# Use 2D case as the main test
ARGS = {
    "X_u": X_u_2d,
    "X_l": X_l_2d
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
kernel_rq Ran
ran; our contract draft was wrong, not the code
rajeevyasarla/ART-SS/ART_SS.py
code served (permissive licence) · get_code("cc9fdeb6fed4e837")
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
import torch.nn as nn

torch.manual_seed(0)

# Define pdist_ker as used in the function
pdist_ker = nn.PairwiseDistance(p=2)

# Define kernel_distance function
def kernel_distance(X_u, X_l):
    xu_shape = X_u.shape
    xl_shape = X_l.shape
    if len(xu_shape) == 2:
        x_l_t = X_l.repeat(xu_shape[0], 1)
        x_u_t = X_u.repeat(xl_shape[0], 1)
        ker_t = pdist_ker(x_u_t, x_l_t)
        ker_t = ker_t.view(xu_shape[0], xl_shape[0])
    elif len(xu_shape) == 3:
        x_l_t = X_l.repeat(1, xu_shape[1], 1)
        x_u_t = X_u.repeat(1, xl_shape[1], 1)
        ker_t = pdist_ker(x_u_t.transpose(1, 2), x_l_t.transpose(1, 2))
        ker_t = ker_t.view(xu_shape[0], xu_shape[1], xl_shape[1])
    elif len(xu_shape) == 4:
        X_u = X_u.reshape(xu_shape[0] * xu_shape[1], xu_shape[2], xu_shape[3])
        X_l = X_l.reshape(xl_shape[0] * xl_shape[1], xl_shape[2], xl_shape[3])
        x_l_t = X_l.repeat(1, xu_shape[2], 1)
        x_u_t = X_u.repeat(1, xl_shape[2], 1)
        ker_t = pdist_ker(x_u_t.transpose(1, 2), x_l_t.transpose(1, 2))
        ker_t = ker_t.view(xu_shape[0], xu_shape[1], xu_shape[2], xl_shape[2])
    return ker_t

# Define kernel_rq function
def kernel_rq(x, y, var, alpha=0.5):
    sigma_1 = 1.0
    pw = 0.6
    l_1 = torch.max(var)
    d = kernel_distance(x, y)
    Ker = sigma_1 ** 2 * (1 + 0.5 * d / (alpha * l_1 ** 2)) ** (-1 * alpha)
    return Ker

# Construct arguments
# x: tensor for unlabeled data - 2D case: shape (2, 4)
x = torch.randn(2, 4)

# y: tensor for labeled data - 2D case: shape (3, 4)
y = torch.randn(3, 4)

# var: variance tensor - needs to be compatible with torch.max()
var = torch.randn(5)

# alpha: scalar parameter (already has default value 0.5)
alpha = 0.5

ARGS = {
    "x": x,
    "y": y,
    "var": var,
    "alpha": alpha
}
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
kernel_rq_np Ran
ran, honoured the contract we drafted
rajeevyasarla/ART-SS/ART_SS.py
code served (permissive licence) · get_code("8954e87a77e94312")
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) and returned ndarray, 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 numpy as np
from scipy.spatial.distance import cdist

# Create small numpy arrays for x and y
# x and y should be 2D arrays where rows are samples
x = np.random.randn(2, 4).astype(np.float32)  # 2 samples, 4 dimensions
y = np.random.randn(3, 4).astype(np.float32)  # 3 samples, 4 dimensions

# Create var array
# The function calls var.max(axis=-1).max(axis=-1).max(axis=-1)
# This suggests var is a 3D or higher dimensional array
# We need to construct it so that three successive max operations work
var = np.random.randn(2, 3, 4).astype(np.float32)  # 3D array

# alpha is a scalar parameter with default 0.5
alpha = 0.5

ARGS = {
    "x": x,
    "y": y,
    "var": var,
    "alpha": alpha
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, scipy — 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 (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
kernel_se Ran
ran; our contract draft was wrong, not the code
rajeevyasarla/ART-SS/ART_SS.py
code served (permissive licence) · get_code("b6007c4da5fa58ce")
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
import torch.nn as nn

torch.manual_seed(0)

# Define pdist_ker as used in the function
pdist_ker = nn.PairwiseDistance(p=2)

def kernel_distance(X_u, X_l):
    xu_shape = X_u.shape
    xl_shape = X_l.shape
    if len(xu_shape) == 2:
        x_l_t = X_l.repeat(xu_shape[0], 1)
        x_u_t = X_u.repeat(xl_shape[0], 1)
        ker_t = pdist_ker(x_u_t, x_l_t)
        ker_t = ker_t.view(xu_shape[0], xl_shape[0])
    elif len(xu_shape) == 3:
        x_l_t = X_l.repeat(1, xu_shape[1], 1)
        x_u_t = X_u.repeat(1, xl_shape[1], 1)
        ker_t = pdist_ker(x_u_t.transpose(1, 2), x_l_t.transpose(1, 2))
        ker_t = ker_t.view(xu_shape[0], xu_shape[1], xl_shape[1])
    elif len(xu_shape) == 4:
        X_u = X_u.reshape(xu_shape[0] * xu_shape[1], xu_shape[2], xu_shape[3])
        X_l = X_l.reshape(xl_shape[0] * xl_shape[1], xl_shape[2], xl_shape[3])
        x_l_t = X_l.repeat(1, xu_shape[2], 1)
        x_u_t = X_u.repeat(1, xl_shape[2], 1)
        ker_t = pdist_ker(x_u_t.transpose(1, 2), x_l_t.transpose(1, 2))
        ker_t = ker_t.view(xu_shape[0], xu_shape[1], xu_shape[2], xl_shape[2])
    return ker_t

def kernel_se(x, y, var):
    sigma_1 = 1.0
    pw = 0.6
    l_1 = torch.max(var)
    d = kernel_distance(x, y)
    Ker = sigma_1 ** 2 * torch.exp(-0.5 * d / l_1 ** 2)
    return Ker

# Create tensors for testing with 2D case
x = torch.randn(2, 4)
y = torch.randn(3, 4)
var = torch.randn(5)

ARGS = {
    'x': x,
    'y': y,
    'var': var
}
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
kernel_se_np Ran
ran, honoured the contract we drafted
rajeevyasarla/ART-SS/ART_SS.py
code served (permissive licence) · get_code("a8eeb196daee4baf")
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) and returned ndarray, 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 numpy as np
from scipy.spatial.distance import cdist

# Create small numpy arrays for x and y
x = np.random.randn(2, 4)  # 2 samples, 4 dimensions
y = np.random.randn(3, 4)  # 3 samples, 4 dimensions

# Create var with shape that supports .max(axis=-1).max(axis=-1).max(axis=-1)
# This means var needs at least 3 dimensions
var = np.random.randn(2, 3, 4)  # 3D array

ARGS = {
    'x': x,
    'y': y,
    'var': var
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, scipy — 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 (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
ART Not yet run
unverified (harvested, no recorded run)
rajeevyasarla/ART-SS/ART_SS.py
code served (permissive licence) · get_code("1e5cb1e88bf6e996")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 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 7 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

In recent years, convolutional neural network-based single image adverse weather removal methods have achieved significant performance improvements on many benchmark datasets. However, these methods require large amounts of clean-weather degraded image pairs for training, which is often difficult to obtain in practice. Although various weather degradation synthesis methods exist in the literature, the use of synthetically generated weather degraded images often results in sub-optimal performance on the real weatherdegraded images due to the domain gap between synthetic and real world images. To deal with this problem, various semi-supervised restoration (SSR) methods have been proposed for deraining or dehazing which learn to restore clean image using synthetically generated datasets while generalizing better using unlabeled real-world images. The performance of a semi-supervised method is essentially based on the quality of the unlabeled data. In particular, if the unlabeled data characteristics are very different from that of the labeled data, then the performance of a semi-supervised method degrades significantly. We theoretically study the effect of unlabeled data on the performance of an SSR method and develop a technique that rejects the unlabeled images that degrade the performance. Extensive experiments and ablation study show that the proposed sample rejection method increases the performance of existing SSR deraining and dehazing methods significantly. Code is available at :https://github.com/rajeevyasarla/ART-SS

For agents

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

get_harvested_code_for_paper("2203.09275")
get_code_for_paper("2203.09275")
have("2203.09275")

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

curl https://syntology.ai/api/ran/2203.09275.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.09275.svg)](https://syntology.ai/paper/2203.09275)

Connect an agent — have() is free.