SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1609.05158 · 2016

Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network

arXiv · PDF · Open in the Atlas

Code that ran

19 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; 1 honoured the contract we drafted, 0 violated it and 13 have no contract verdict; 5 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/1609.05158.json · badge: /api/ran/1609.05158.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
niazwazir/REAL_TIME_VIDEO_IMSR community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
Nhat-Thanh/ESPCN-TF community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
Nhat-Thanh/ESPCN-Pytorch community (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
npielawski/pytorch_tiramisu community (classifier label)
classifier basis not recorded on the link
2 ran · 0 no recorded run
One-sixth/pixelshuffle_invert_pytorch extension (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
HighVoltageRocknRoll/sr community (classifier label)
classifier basis not recorded on the link
1 ran · 1 no recorded run
behnoudshafizadeh/ESPCN-SuperResolution-Tensorflow community (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
Lornatang/ESPCN-PyTorch reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
PSNR Ran
ran, honoured the contract we drafted
behnoudshafizadeh/ESPCN-SuperResolution-Tensorflow/utils.py
code served (permissive licence) · get_code("5a900d242c89ecf1")
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
import math

torch_imported = False
try:
    import torch
    torch_imported = True
except:
    pass

# The PSNR function takes two array-like arguments
# It compares orig and reconstr by computing MSE and then PSNR
# Both need to be convertible to float via .astype(float)
# They should have the same shape for comparison to make sense

# Create small numpy arrays with distinct dimensions to probe shapes
orig = np.array([[[[10, 20, 30, 40],
                    [50, 60, 70, 80],
                    [90, 100, 110, 120]],
                   [[130, 140, 150, 160],
                    [170, 180, 190, 200],
                    [210, 220, 230, 240]]]], dtype=np.uint8)

# reconstr should have same shape but different values
reconstr = np.array([[[[12, 22, 32, 42],
                       [52, 62, 72, 82],
                       [92, 102, 112, 122]],
                      [[132, 142, 152, 162],
                       [172, 182, 192, 202],
                       [212, 222, 232, 242]]]], dtype=np.uint8)

ARGS = {
    "orig": orig,
    "reconstr": reconstr
}
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 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
bytes_feature Ran
ran; our contract draft was wrong, not the code
HighVoltageRocknRoll/sr/datasets/prepare_dataset.py
code served (permissive licence) · get_code("d682a9593710857e")
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 Feature. 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 tensorflow as tf

# bytes_feature expects a single parameter 'value' which should be bytes
# The function wraps it in tf.train.BytesList and then tf.train.Feature

ARGS = {
    "value": b"example_bytes_data"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Feature — a Feature output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Feature 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
calculate_valid_crop_size Ran
ran on a synthesized input
niazwazir/REAL_TIME_VIDEO_IMSR/data_utils.py
code served (permissive licence) · get_code("c889b417f65ffc2a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

crop_size = 128
upscale_factor = 4

ARGS = {
    "crop_size": crop_size,
    "upscale_factor": upscale_factor
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
center_crop Ran
ran on a synthesized input
npielawski/pytorch_tiramisu/tiramisu/tiramisu.py
code served (permissive licence) · get_code("fbf4fd83b502b493")
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, 4, 4, 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)

# Create a small 4D tensor (batch, channels, height, width)
# Using distinct sizes: batch=2, channels=4, height=8, width=6
layer = torch.randn(2, 4, 8, 6)

# Define crop dimensions (smaller than the layer dimensions)
max_height = 4
max_width = 3

ARGS = {
    "layer": layer,
    "max_height": max_height,
    "max_width": max_width,
}
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, 4, 4, 3) — shape (2, 4, 4, 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
choice_device Ran
ran; our contract draft was wrong, not the code
Lornatang/ESPCN-PyTorch/inference.py
code served (permissive licence) · get_code("3ddf550638922b31")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned device. 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 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 device_type string and returns a torch.device
# We should pass 'cpu' since we cannot use 'cuda' in this environment
device_type = 'cpu'

ARGS = {
    'device_type': device_type
}
Setup block drafted by
a language model, Claude Sonnet 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 device — a device output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a device 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
crop_asym_like Ran
ran on a synthesized input
npielawski/pytorch_tiramisu/tiramisu/tiramisu.py
code served (permissive licence) · get_code("559b2398ad70b2dd")
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, 4, 5, 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)

# Create two 4D tensors (batch, channels, height, width)
# layer tensor should be larger than or equal to like tensor
layer = torch.randn(2, 4, 8, 6)  # batch=2, channels=4, height=8, width=6
like = torch.randn(2, 4, 5, 3)   # batch=2, channels=4, height=5, width=3

ARGS = {
    "layer": layer,
    "like": like
}
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, 4, 5, 3) — shape (2, 4, 5, 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
extend Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from alexjc/neural-enhance
pointer only; the stored copy get_code serves, from alexjc/neural-enhance, is pointer only (licence: AGPL-3.0) · get_code("a8dabd6fc5e7e33d")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

is_image_file Ran
ran on a synthesized input
niazwazir/REAL_TIME_VIDEO_IMSR/data_utils.py
code served (permissive licence) · get_code("7e6cb8a85c64996e")
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
import torch.nn as nn

ARGS = {
    'filename': 'test_image.png'
}
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
is_video_file Ran
ran on a synthesized input
niazwazir/REAL_TIME_VIDEO_IMSR/data_utils.py
code served (permissive licence) · get_code("b8a23b27c781a66c")
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
import numpy as np

def is_video_file(filename):
    return any((filename.endswith(extension) for extension in ['.mp4', '.avi', '.mpg', '.mkv', '.wmv', '.flv']))

ARGS = {
    'filename': 'test_video.mp4'
}
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 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
pixel_shuffle Ran
ran on a synthesized input
Nhat-Thanh/ESPCN-TF/neuralnet.py
code served (permissive licence) · get_code("c8d3363d1625adc6")
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 function. 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 tensorflow as tf
import numpy as np

def pixel_shuffle(scale):
    return lambda x: tf.nn.depth_to_space(x, scale)

# Create a small test tensor
# depth_to_space expects input shape: [batch, height, width, depth]
# We'll create a tensor with batch=2, height=2, width=2, depth=16
# This allows scale=2 to work (depth must be divisible by scale^2)
input_tensor = tf.constant(np.random.randn(2, 2, 2, 16).astype(np.float32))

# Test the function
scale = 2

ARGS = {
    "scale": scale
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
pixelshuffle Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
One-sixth/pixelshuffle_invert_pytorch/pixelshuffle_invert.py
code served (permissive licence) · get_code("8c2feff80635400e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned Tensor, ndim 4, dtype family float, shape (2, 2, 8, 15) 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 performs pixel shuffle operation
# It takes a tensor and upsamples it by rearranging channels into spatial dimensions
# 
# Input shape: (B, iC, iH, iW)
# Output shape: (B, oC, oH, oW) where:
#   oC = iC // (pH * pW)
#   oH = iH * pH
#   oW = iW * pW
#
# For this to work, iC must be divisible by (pH * pW)

# Let's use factor_hw = (2, 3), so pH=2, pW=3, and pH*pW=6
# We need iC to be divisible by 6
# Let's use iC = 12, so oC = 12 // 6 = 2

B = 2
iC = 12
iH = 4
iW = 5

x = torch.randn(B, iC, iH, iW)
factor_hw = (2, 3)

ARGS = {
    'x': x,
    'factor_hw': factor_hw
}
Setup block drafted by
a language model, Claude Sonnet 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 4, dtype family float, shape (2, 2, 8, 15) — shape (2, 2, 8, 15) 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
pixelshuffle_invert Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
One-sixth/pixelshuffle_invert_pytorch/pixelshuffle_invert.py
code served (permissive licence) · get_code("85cdae1b9119a1aa")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned Tensor, ndim 4, dtype family float, shape (2, 18, 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 performs an inverse pixel shuffle operation
# It takes a tensor and reshapes it by a factor (pH, pW)
# Input shape: (B, iC, iH, iW)
# Output shape: (B, iC * pH * pW, iH // pH, iW // pW)
# So iH must be divisible by pH and iW must be divisible by pW

B = 2
iC = 3
pH = 2
pW = 3
iH = 4  # divisible by pH=2
iW = 6  # divisible by pW=3

x = torch.randn(B, iC, iH, iW)
factor_hw = (pH, pW)

ARGS = {
    'x': x,
    'factor_hw': factor_hw
}
Setup block drafted by
a language model, Claude Sonnet 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 4, dtype family float, shape (2, 18, 2, 2) — shape (2, 18, 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
psnr Ran
ran on a synthesized input
niazwazir/REAL_TIME_VIDEO_IMSR/PSNR.py
code served (permissive licence) · get_code("440d24ec30b09c77")
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
import torch

torch.manual_seed(0)

# The function takes two images and computes PSNR
# It performs arithmetic operations: subtraction, squaring, mean
# The inputs should be numpy arrays or similar that support these operations
# The values should be in the range [0, 255] for typical image data

# Create two small image arrays (2x4 for distinct dimensions)
img1 = np.array([[100, 150, 200, 50],
                  [75, 125, 175, 225]], dtype=np.float32)

img2 = np.array([[102, 148, 198, 52],
                  [73, 127, 177, 223]], dtype=np.float32)

ARGS = {
    "img1": img1,
    "img2": img2
}
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 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
read_image Ran
ran on a synthesized input
Nhat-Thanh/ESPCN-Pytorch/utils/common.py
code served (permissive licence) · get_code("4d448d32e15f04c9")
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 int, shape (3, 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 torchvision.io as io
from PIL import Image
import numpy as np
import tempfile
import os

# Create a temporary image file for testing
torch.manual_seed(0)

# Create a simple RGB image using PIL
temp_dir = tempfile.mkdtemp()
img_array = np.random.randint(0, 256, (4, 3, 3), dtype=np.uint8)
img = Image.fromarray(img_array.transpose(1, 2, 0), mode='RGB')
filepath = os.path.join(temp_dir, 'test_image.png')
img.save(filepath)

ARGS = {
    'filepath': filepath
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, torch, torchvision — 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 int, shape (3, 3, 3) — shape (3, 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
read_image Ran
ran on a synthesized input
Nhat-Thanh/ESPCN-TF/utils/common.py
code served (permissive licence) · get_code("05f12c47fb9b8978")
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 EagerTensor, ndim 3, dtype family int, shape (2, 4, 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 tensorflow as tf
import tempfile
import os

# Create a temporary directory and a small test image file
temp_dir = tempfile.mkdtemp()
filepath = os.path.join(temp_dir, "test_image.png")

# Create a small 4x3 RGB image and save it
image_array = tf.constant([[[255, 0, 0], [0, 255, 0], [0, 0, 255], [255, 255, 0]],
                           [[128, 128, 128], [64, 64, 64], [32, 32, 32], [200, 100, 50]]], dtype=tf.uint8)
tf.io.write_file(filepath, tf.io.encode_png(image_array))

ARGS = {
    "filepath": filepath
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 3, dtype family int, shape (2, 4, 3) — shape (2, 4, 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
rgb2ycbcr Ran
ran on a synthesized input
Nhat-Thanh/ESPCN-Pytorch/utils/common.py
code served (permissive licence) · get_code("d8058eded6a9e999")
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 int, shape (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

torch.manual_seed(0)

# The function expects a tensor with shape (3, height, width)
# where the first dimension represents R, G, B channels
# The function indexes into dimension 0 to get individual channels
# and performs operations on them. We need a small tensor with values
# in a typical image range (0-255)

src = torch.randint(0, 256, (3, 4, 4), dtype=torch.float32)

ARGS = {
    "src": src
}
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 int, shape (3, 4, 4) — shape (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
rgb2ycbcr Ran
ran on a synthesized input
Nhat-Thanh/ESPCN-TF/utils/common.py
code served (permissive licence) · get_code("3760ee509acdf19f")
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 EagerTensor, ndim 3, dtype family int, shape (4, 6, 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
import tensorflow as tf

# Create a small RGB image with shape (height, width, channels)
# Using small dimensions: 4 height, 6 width, 3 RGB channels
rgb_image = np.random.randint(0, 256, size=(4, 6, 3), dtype=np.uint8)

ARGS = {
    "src": rgb_image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 3, dtype family int, shape (4, 6, 3) — shape (4, 6, 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
ycbcr2rgb Ran
ran on a synthesized input
Nhat-Thanh/ESPCN-Pytorch/utils/common.py
code served (permissive licence) · get_code("ef42ea280e97a9f8")
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 int, shape (3, 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)

# The function expects src to be a tensor with shape (3, height, width)
# where src[0] is Y channel, src[1] is Cb channel, src[2] is Cr channel
# Create a small YCbCr tensor with shape (3, 2, 4)
src = torch.rand(3, 2, 4) * 256

ARGS = {
    'src': src
}
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 int, shape (3, 2, 4) — shape (3, 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
ycbcr2rgb Ran
ran on a synthesized input
Nhat-Thanh/ESPCN-TF/utils/common.py
code served (permissive licence) · get_code("3b0edbcfb724a432")
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 EagerTensor, 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 numpy as np
import tensorflow as tf

# Create a small YCbCr image array
# Shape: (height=4, width=3, channels=3) for YCbCr
ycbcr_image = np.random.randint(0, 256, size=(4, 3, 3), dtype=np.uint8).astype(np.float32)

# Ensure values are in reasonable YCbCr ranges
# Y channel: 0-255, Cb/Cr: typically 0-255 (with 128 as neutral)
ycbcr_image[:, :, 0] = np.random.uniform(16, 235, size=(4, 3))  # Y range
ycbcr_image[:, :, 1] = np.random.uniform(16, 240, size=(4, 3))  # Cb range
ycbcr_image[:, :, 2] = np.random.uniform(16, 240, size=(4, 3))  # Cr range

ARGS = {
    "src": ycbcr_image
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
image_warp Not yet run
unverified (harvested, no recorded run)
HighVoltageRocknRoll/sr/models/image_warp.py
code served (permissive licence) · get_code("307b25262433fb97")
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, 12 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 3 the same month, 4 later) and 8 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
bytes_feature 8 other papers
2 dated earlier, 6 later
1409.1556 (2014-09)
Very Deep Convolutional Networks for Large-Scale Image Recognition
1607.08022 (2016-07)
Instance Normalization: The Missing Ingredient for Fast Stylization
1802.05591 (2018-02)
Towards End-to-End Lane Detection: an Instance Segmentation Approach
and 5 more not named here
calculate_valid_crop_size 3 other papers
1 the same month, 2 later
1609.04802 (2016-09)
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
1802.08797 (2018-02)
Residual Dense Network for Image Super-Resolution
1911.07850 (2019-11)
Frequency Separation for Real-World Super-Resolution
read_image 3 other papers
3 dated earlier
1501.00092 (2015-01)
Image Super-Resolution Using Deep Convolutional Networks
1511.04587 (2015-11)
Accurate Image Super-Resolution Using Very Deep Convolutional Networks
1608.00367 (2016-08)
Accelerating the Super-Resolution Convolutional Neural Network
rgb2ycbcr 3 other papers
3 dated earlier
1501.00092 (2015-01)
Image Super-Resolution Using Deep Convolutional Networks
1511.04587 (2015-11)
Accurate Image Super-Resolution Using Very Deep Convolutional Networks
1608.00367 (2016-08)
Accelerating the Super-Resolution Convolutional Neural Network
ycbcr2rgb 3 other papers
3 dated earlier
1501.00092 (2015-01)
Image Super-Resolution Using Deep Convolutional Networks
1511.04587 (2015-11)
Accurate Image Super-Resolution Using Very Deep Convolutional Networks
1608.00367 (2016-08)
Accelerating the Super-Resolution Convolutional Neural Network
read_image 2 other papers
2 dated earlier
1501.00092 (2015-01)
Image Super-Resolution Using Deep Convolutional Networks
1608.00367 (2016-08)
Accelerating the Super-Resolution Convolutional Neural Network
rgb2ycbcr 2 other papers
2 dated earlier
1501.00092 (2015-01)
Image Super-Resolution Using Deep Convolutional Networks
1608.00367 (2016-08)
Accelerating the Super-Resolution Convolutional Neural Network
ycbcr2rgb 2 other papers
2 dated earlier
1501.00092 (2015-01)
Image Super-Resolution Using Deep Convolutional Networks
1608.00367 (2016-08)
Accelerating the Super-Resolution Convolutional Neural Network
PSNR 1 other paper
1 later
1710.01992 (2017-10)
Fast and Accurate Image Super-Resolution with Deep Laplacian Pyramid Networks
choice_device 1 other paper
1 later
1903.00875 (2019-03)
Meta-SR: A Magnification-Arbitrary Network for Super-Resolution
extend 1 other paper
1 the same month
1609.04802 (2016-09)
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
pixel_shuffle 1 other paper
1 the same month
1609.04802 (2016-09)
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

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

More are linked than shown here.

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

Abstract

Recently, several models based on deep neural networks have achieved great success in terms of both reconstruction accuracy and computational performance for single image super-resolution. In these methods, the low resolution (LR) input image is upscaled to the high resolution (HR) space using a single filter, commonly bicubic interpolation, before reconstruction. This means that the super-resolution (SR) operation is performed in HR space. We demonstrate that this is sub-optimal and adds computational complexity. In this paper, we present the first convolutional neural network (CNN) capable of real-time SR of 1080p videos on a single K2 GPU. To achieve this, we propose a novel CNN architecture where the feature maps are extracted in the LR space. In addition, we introduce an efficient sub-pixel convolution layer which learns an array of upscaling filters to upscale the final LR feature maps into the HR output. By doing so, we effectively replace the handcrafted bicubic filter in the SR pipeline with more complex upscaling filters specifically trained for each feature map, whilst also reducing the computational complexity of the overall SR operation. We evaluate the proposed approach using images and videos from publicly available datasets and show that it performs significantly better (+0.15dB on Images and +0.39dB on Videos) and is an order of magnitude faster than previous CNN-based methods.

For agents

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

get_harvested_code_for_paper("1609.05158")
get_code_for_paper("1609.05158")
have("1609.05158")

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

curl https://syntology.ai/api/ran/1609.05158.json

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

[![Syntology run record](https://syntology.ai/api/ran/1609.05158.svg)](https://syntology.ai/paper/1609.05158)

Connect an agent — have() is free.