SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1808.02473 · 2018

SketchyScene: Richly-Annotated Scene Sketches

arXiv · PDF · Open in the Atlas

Code that ran

10 of 11 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 1 violated it and 7 have no contract verdict; 1 failure 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/1808.02473.json · badge: /api/ran/1808.02473.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
SketchyScene/SketchyScene canonical (classifier label)
classifier basis: census
10 ran · 1 no recorded run
FunctionStatusWhere it lives
compute_iou Ran
ran, violated the contract we drafted
SketchyScene/SketchyScene/Instance_Segmentation/libs/utils.py
code served (permissive licence) · get_code("b2aae18de5fa3893")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (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 numpy as np

# Create a single box [y1, x1, y2, x2]
box = np.array([10.0, 20.0, 30.0, 40.0])

# Create multiple boxes for comparison
boxes = np.array([
    [12.0, 22.0, 28.0, 38.0],
    [5.0, 15.0, 35.0, 45.0]
])

# Calculate box area: (y2 - y1) * (x2 - x1)
box_area = (box[2] - box[0]) * (box[3] - box[1])

# Calculate areas for each box in boxes
boxes_area = np.array([
    (boxes[0, 2] - boxes[0, 0]) * (boxes[0, 3] - boxes[0, 1]),
    (boxes[1, 2] - boxes[1, 0]) * (boxes[1, 3] - boxes[1, 1])
])

ARGS = {
    "box": box,
    "boxes": boxes,
    "box_area": box_area,
    "boxes_area": boxes_area
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (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
compute_overlaps Ran
ran, honoured the contract we drafted
SketchyScene/SketchyScene/Instance_Segmentation/libs/utils.py
code served (permissive licence) · get_code("698f3241ee39ad75")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (3, 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 numpy as np

# Create sample bounding boxes in format [y1, x1, y2, x2]
# boxes1: 3 boxes
boxes1 = np.array([
    [10, 20, 50, 80],
    [30, 40, 70, 100],
    [5, 5, 25, 35]
], dtype=np.float32)

# boxes2: 2 boxes
boxes2 = np.array([
    [15, 25, 60, 90],
    [25, 35, 65, 95]
], dtype=np.float32)

ARGS = {
    "boxes1": boxes1,
    "boxes2": boxes2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (3, 2) — shape (3, 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
conv_block Ran
ran on a synthesized input
SketchyScene/SketchyScene/Instance_Segmentation/libs/model.py
code served (permissive licence) · get_code("f998b7fc598210b3")
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 KerasTensor. 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 keras.layers as KL
import keras
import numpy as np

# Create a small input tensor for the conv_block function
# Shape: (batch_size, height, width, channels)
input_tensor = keras.Input(shape=(8, 8, 3))

ARGS = {
    'input_tensor': input_tensor,
    'kernel_size': 3,
    'filters': [64, 64, 256],
    'stage': 3,
    'block': 'a',
    'strides': (2, 2),
    'use_bias': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
keras, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type KerasTensor — a KerasTensor output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
expand_small_segmentation_mask Ran
ran on a synthesized input
SketchyScene/SketchyScene/Instance_Segmentation/segment_data_generation.py
code served (permissive licence) · get_code("d5e134d3e4c917dd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family int, shape (2, 768, 768) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create small segmentation masks (list of 2 small masks)
pred_masks_small_list = [
    np.random.randint(0, 2, size=(32, 32), dtype=np.uint8),
    np.random.randint(0, 2, size=(32, 32), dtype=np.uint8),
]

# Create bounding boxes for each mask [y1, x1, y2, x2]
# Use distinct coordinates for each box
pred_boxes = [
    [100, 150, 131, 181],  # First box: y1=100, x1=150, y2=131, x2=181
    [250, 300, 281, 331],  # Second box: y1=250, x1=300, y2=281, x2=331
]

ARGS = {
    "pred_masks_small_list": pred_masks_small_list,
    "pred_boxes": pred_boxes,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (2, 768, 768) — shape (2, 768, 768) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
extract_bboxes Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
SketchyScene/SketchyScene/Instance_Segmentation/libs/utils.py
code served (permissive licence) · get_code("d2c9b72e23d9588a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family int, shape (3, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a simple test mask with shape [height, width, num_instances]
# Let's use height=5, width=6, num_instances=3
height = 5
width = 6
num_instances = 3

mask = np.zeros((height, width, num_instances), dtype=np.uint8)

# Instance 0: small rectangle in top-left
mask[1:3, 1:3, 0] = 1

# Instance 1: small rectangle in bottom-right
mask[3:5, 4:6, 1] = 1

# Instance 2: single pixel
mask[2, 3, 2] = 1

ARGS = {
    "mask": mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family int, shape (3, 4) — shape (3, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_edgelist Ran
ran on a synthesized input
SketchyScene/SketchyScene/Instance_Segmentation/libs/edgelist_utils.py
code served (permissive licence) · get_code("955c97524949f532")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (8, 8) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
import tempfile
import numpy as np
from PIL import Image
import scipy.io

# Create temporary directories for the test
temp_dir = tempfile.mkdtemp()
sketchyscene_base_dir = os.path.join(temp_dir, 'sketchyscene')
edgelist_base_dir = os.path.join(temp_dir, 'edgelists')

# Create directory structure
os.makedirs(os.path.join(sketchyscene_base_dir, 'test', 'DRAWING_GT'), exist_ok=True)
os.makedirs(os.path.join(sketchyscene_base_dir, 'val', 'DRAWING_GT'), exist_ok=True)
os.makedirs(os.path.join(edgelist_base_dir, 'test'), exist_ok=True)
os.makedirs(os.path.join(edgelist_base_dir, 'val'), exist_ok=True)

# Create a small test image (8x8 RGB image with some black pixels for sketch)
image_id = 1
height, width = 8, 8
sketch_array = np.ones((height, width, 3), dtype=np.uint8) * 255  # White background
# Add some black pixels (sketch lines)
sketch_array[2:4, 2:6, :] = 0
sketch_array[4:6, 3:5, :] = 0

# Save the sketch image
sketch_image_path = os.path.join(sketchyscene_base_dir, 'test', 'DRAWING_GT', f'L0_sample{image_id}.png')
Image.fromarray(sketch_array, 'RGB').save(sketch_image_path)

# Create edgelist mat file with labeled edges
edgelist_array = np.zeros((height, width), dtype=np.float32)
# Add some edge labels near the sketch
edgelist_array[1, 2:6] = 1.0
edgelist_array[3:5, 1] = 1.0
edgelist_array[5, 3:5] = 2.0

edgelist_mat_path = os.path.join(edgelist_base_dir, 'test', f'edgelist_{image_id}.mat')
scipy.io.savemat(edgelist_mat_path, {'labelededgeim': edgelist_array})

ARGS = {
    'image_id': 1,
    'data_type': 'test',
    'sketchyscene_base_dir': sketchyscene_base_dir,
    'edgelist_base_dir': edgelist_base_dir,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, 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 (8, 8) — shape (8, 8) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
identity_block Ran
ran on a synthesized input
SketchyScene/SketchyScene/Instance_Segmentation/libs/model.py
code served (permissive licence) · get_code("572ccdbc1739d878")
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 KerasTensor. 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 keras.layers as KL
import numpy as np

# Create a small input tensor for the identity block
# Shape: (batch_size=2, height=4, width=4, channels=64)
input_tensor = KL.Input(shape=(4, 4, 64))

# Define parameters
kernel_size = 3
filters = [64, 64, 64]  # nb_filter1, nb_filter2, nb_filter3
stage = 2
block = 'a'
use_bias = True

ARGS = {
    'input_tensor': input_tensor,
    'kernel_size': kernel_size,
    'filters': filters,
    'stage': stage,
    'block': block,
    'use_bias': use_bias,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
keras, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type KerasTensor — a KerasTensor output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
label2d_array_nn_scaling Ran
ran on a synthesized input
SketchyScene/SketchyScene/Instance_Segmentation/libs/edgelist_utils.py
code served (permissive licence) · get_code("60c04ed4c9d8ff7a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family int, shape (8, 6) 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 scipy.ndimage

# Create a small 2D label array (4x3)
label2d = np.array([
    [0, 1, 2],
    [1, 2, 0],
    [2, 0, 1],
    [0, 1, 2]
], dtype=np.int32)

# Define new dimensions (distinct from original)
new_h = 8
new_w = 6

ARGS = {
    "label2d": label2d,
    "new_h": new_h,
    "new_w": new_w,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 int, shape (8, 6) — shape (8, 6) 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
myConvertFun Ran
ran on a synthesized input
SketchyScene/SketchyScene/Semantic_Segmentation/adapted_deeplab_model.py
code served (permissive licence) · get_code("36da9f33aed27e8d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 4, dtype family float, shape (2, 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 numpy as np

# Create a small 4D numpy array
# Shape: (batch=2, channels=3, height=4, width=5)
# The function slices [:, :, :, 1:] which removes the first element of the last dimension
# Then squeezes to remove dimensions of size 1
tensor_input = np.random.randn(2, 3, 4, 5).astype(np.float32)

ARGS = {
    "tensor_": tensor_input
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 4, dtype family float, shape (2, 3, 4, 4) — shape (2, 3, 4, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
resnet_graph Ran
ran on a synthesized input
SketchyScene/SketchyScene/Instance_Segmentation/libs/model.py
code served (permissive licence) · get_code("5883d43b9e07bf2e")
How it ran

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

import keras.layers as KL
import tensorflow as tf
import numpy as np

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

# Create a small input tensor (batch=2, height=32, width=32, channels=3)
input_image = KL.Input(shape=(32, 32, 3), batch_size=2)

ARGS = {
    'input_image': input_image,
    'architecture': 'resnet50',
    'stage5': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
keras, numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list — a list output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
refine_mask_with_edgelist Not yet run
unverified (harvested, no recorded run)
SketchyScene/SketchyScene/Instance_Segmentation/libs/edgelist_utils.py
code served (permissive licence) · get_code("f0a92532555ad369")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 3 later) and 5 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
compute_iou 9 other papers
3 dated earlier, 6 later
1612.03242 (2016-12)
StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks
1702.01983 (2017-02)
Face Aging With Conditional Generative Adversarial Networks
and 6 more not named here
compute_overlaps 9 other papers
3 dated earlier, 6 later
1612.03242 (2016-12)
StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks
1702.01983 (2017-02)
Face Aging With Conditional Generative Adversarial Networks
and 6 more not named here
extract_bboxes 9 other papers
3 dated earlier, 6 later
1612.03242 (2016-12)
StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks
1702.01983 (2017-02)
Face Aging With Conditional Generative Adversarial Networks
and 6 more not named here
conv_block 1 other paper
1 dated earlier
1703.06870 (2017-03)
Mask R-CNN
identity_block 1 other paper
1 dated earlier
1703.06870 (2017-03)
Mask R-CNN
resnet_graph 1 other paper
1 dated earlier
1703.06870 (2017-03)
Mask R-CNN

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

We contribute the first large-scale dataset of scene sketches, SketchyScene, with the goal of advancing research on sketch understanding at both the object and scene level. The dataset is created through a novel and carefully designed crowdsourcing pipeline, enabling users to efficiently generate large quantities of realistic and diverse scene sketches. SketchyScene contains more than 29,000 scene-level sketches, 7,000+ pairs of scene templates and photos, and 11,000+ object sketches. All objects in the scene sketches have ground-truth semantic and instance masks. The dataset is also highly scalable and extensible, easily allowing augmenting and/or changing scene composition. We demonstrate the potential impact of SketchyScene by training new computational models for semantic segmentation of scene sketches and showing how the new dataset enables several applications including image retrieval, sketch colorization, editing, and captioning, etc. The dataset and code can be found at https://github.com/SketchyScene/SketchyScene.

For agents

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

get_harvested_code_for_paper("1808.02473")
get_code_for_paper("1808.02473")
have("1808.02473")

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

curl https://syntology.ai/api/ran/1808.02473.json

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

[![Syntology run record](https://syntology.ai/api/ran/1808.02473.svg)](https://syntology.ai/paper/1808.02473)

Connect an agent — have() is free.