SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2005.13899 · 2020

Deep Learning for Automatic Pneumonia Detection

arXiv · PDF · Open in the Atlas

Code that ran

10 of 10 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, 1 violated it and 7 have no contract verdict; 2 failures of our instrument; 0 no recorded run), measured 2026-11-21; 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/2005.13899.json · badge: /api/ran/2005.13899.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
tatigabru/kaggle-rsna canonical (classifier label)
classifier basis: census
10 ran · 0 no recorded run
FunctionStatusWhere it lives
anchors_for_shape Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
tatigabru/kaggle-rsna/src/pytorch_retinanet/anchors.py
code served (permissive licence) · get_code("2a214da3b52ccc74")
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 (18144, 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

ARGS = {
    'image_shape': (256, 384, 3),
    'pyramid_levels': [3, 4, 5],
    'ratios': np.array([0.5, 1, 2]),
    'scales': np.array([2 ** 0, 2 ** (1.0 / 3.0), 2 ** (2.0 / 3.0)]),
    'strides': [8, 16, 32],
    'sizes': [16, 32, 64],
    'shapes_callback': None,
}
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 (18144, 4) — shape (18144, 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
compute_shape Ran
ran, violated the contract we drafted
tatigabru/kaggle-rsna/src/pytorch_retinanet/anchors.py
code served (permissive licence) · get_code("0e14588c12c9c1b4")
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 2, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Analyze the function:
# - image_shape: used as image_shape[:2], then converted to np.array, so needs to be array-like with at least 2 elements
# - pyramid_levels: used in a list comprehension with "for x in pyramid_levels", so needs to be iterable of numbers

# Create distinct sizes for clarity
image_shape = (480, 640, 3)  # 3-element tuple (height, width, channels)
pyramid_levels = [0, 1, 2]   # list of integers representing pyramid levels

ARGS = {
    "image_shape": image_shape,
    "pyramid_levels": pyramid_levels,
}
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 list, ndim 2, dtype family int — 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
create_folds Ran
ran on a synthesized input
tatigabru/kaggle-rsna/src/folds/split_folds.py
code served (permissive licence) · get_code("f4dccf2c3d24f63a")
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 DataFrame. 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 os
import pandas as pd
import numpy as np
from sklearn.model_selection import StratifiedKFold

# Create small sample data for testing
np.random.seed(42)
n_samples = 10

# Create a simple dataframe with an index
df = pd.DataFrame({
    'id': range(n_samples),
    'value': np.random.randn(n_samples)
})

# Create X as a Series (features for stratification)
X = pd.Series(np.random.randn(n_samples), index=range(n_samples))

# Create y as a Series (target for stratification - binary classification)
y = pd.Series(np.random.randint(0, 2, n_samples), index=range(n_samples))

# Define arguments
ARGS = {
    'df': df,
    'X': X,
    'y': y,
    'nb_folds': 3,
    'if_save': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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
generate_anchors Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
tatigabru/kaggle-rsna/src/pytorch_retinanet/anchors.py
code served (permissive licence) · get_code("1279986e0f62b111")
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 (9, 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

ARGS = {
    "base_size": 16,
    "ratios": np.array([0.5, 1, 2]),
    "scales": np.array([2 ** 0, 2 ** (1.0 / 3.0), 2 ** (2.0 / 3.0)])
}
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 (9, 4) — shape (9, 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
iou Ran
ran on a synthesized input
tatigabru/kaggle-rsna/src/metric.py
code served (permissive licence) · get_code("2b7d810a1a5bc468")
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 torch
import numpy as np

# The iou function takes two bounding boxes in format [x, y, w, h]
# where x, y is the top-left corner, w is width, h is height
# Both should be lists of 4 integers

box1 = [10, 20, 50, 60]  # x=10, y=20, width=50, height=60
box2 = [30, 40, 40, 50]  # x=30, y=40, width=40, height=50

ARGS = {
    "box1": box1,
    "box2": box2,
}
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
load_model Ran
ran on a synthesized input
tatigabru/kaggle-rsna/src/predict.py
code served (permissive licence) · get_code("e71f23abf96a7802")
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 Sequential. 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
from torch import nn
import tempfile
import os

torch.manual_seed(0)

# Create a simple model to save
simple_model = nn.Sequential(
    nn.Linear(4, 8),
    nn.ReLU(),
    nn.Linear(8, 3)
)

# Save the model to a temporary file
with tempfile.TemporaryDirectory() as tmpdir:
    checkpoint_path = os.path.join(tmpdir, 'model.pt')
    torch.save(simple_model, checkpoint_path)
    
    # Read the checkpoint path for use in ARGS
    with open(checkpoint_path, 'rb') as f:
        checkpoint_data = f.read()
    
    # Create a persistent temporary file that won't be deleted
    final_checkpoint = tempfile.NamedTemporaryFile(delete=False, suffix='.pt')
    final_checkpoint.write(checkpoint_data)
    final_checkpoint.close()
    checkpoint_file = final_checkpoint.name

ARGS = {
    'checkpoint': checkpoint_file
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
load_oof Ran
ran on a synthesized input
tatigabru/kaggle-rsna/src/scores.py
code served (permissive licence) · get_code("d46a2af9eb5256b6")
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 os
import pickle
import numpy as np
import tempfile

# Create a temporary directory structure that mimics the expected results_dir layout
temp_dir = tempfile.mkdtemp()
results_dir = temp_dir
model_name = "test_model"
fold = 0
epoch_num = 1

# Create the directory structure
oof_dir = os.path.join(results_dir, "oof", f"{model_name}_fold_{fold}")
os.makedirs(oof_dir, exist_ok=True)

# Create a sample out-of-fold predictions array and save it as pickle
sample_oof = np.array([[0.1, 0.9], [0.3, 0.7], [0.2, 0.8]], dtype=np.float32)
pickle_file = os.path.join(oof_dir, f"{epoch_num:03}.pkl")
with open(pickle_file, 'wb') as f:
    pickle.dump(sample_oof, f)

ARGS = {
    "results_dir": results_dir,
    "model_name": model_name,
    "fold": fold,
    "epoch_num": epoch_num,
}
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
map_iou Ran
ran on a synthesized input
tatigabru/kaggle-rsna/src/metric.py
code served (permissive licence) · get_code("43462ea45ba2ce40")
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

# Construct small, valid test cases for map_iou function
# boxes format: (x, y, w, h) where x,y is top-left corner, w,h are width/height

# Ground truth boxes: 2 boxes
boxes_true = np.array([
    [10, 10, 50, 50],   # box 1: top-left (10,10), width 50, height 50
    [100, 100, 40, 40]  # box 2: top-left (100,100), width 40, height 40
], dtype=np.float32)

# Predicted boxes: 3 boxes
boxes_pred = np.array([
    [12, 12, 48, 48],   # box 1: slightly offset from true box 1
    [105, 105, 35, 35], # box 2: slightly offset from true box 2
    [200, 200, 30, 30]  # box 3: false positive
], dtype=np.float32)

# Scores for each predicted box (in any order, will be sorted by function)
scores = np.array([0.9, 0.8, 0.7], dtype=np.float32)

# IoU thresholds to evaluate
thresholds = (0.4, 0.5, 0.6, 0.7)

ARGS = {
    'boxes_true': boxes_true,
    'boxes_pred': boxes_pred,
    'scores': scores,
    'thresholds': thresholds
}
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 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
p1p2_to_xywh Ran
ran on a synthesized input
tatigabru/kaggle-rsna/src/scores.py
code served (permissive licence) · get_code("bdfb5aa233eeea49")
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 (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 numpy as np

# Create a small numpy array with shape (2, 4)
# where first 2 columns are p1 (x0, y0) and last 2 columns are p2 (x1, y1)
p1p2 = np.array([
    [10.0, 20.0, 50.0, 80.0],  # box 1: top-left (10, 20), bottom-right (50, 80)
    [5.0, 15.0, 35.0, 65.0]    # box 2: top-left (5, 15), bottom-right (35, 65)
])

ARGS = {
    "p1p2": p1p2
}
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 (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
validation Ran
ran on a synthesized input
tatigabru/kaggle-rsna/src/train_runner.py
code served (permissive licence) · get_code("7286ed34a1332306")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. 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.nn as nn
import tempfile
import os

torch.manual_seed(0)

class MockRetinaNet(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv = nn.Conv2d(3, 16, 3)
    
    def forward(self, inputs, return_loss=False, return_boxes=False):
        img, annot, category = inputs
        batch_size = img.shape[0]
        
        # Return format: (classification_loss, regression_loss, global_classification_loss, 
        #                  nms_scores, global_class, transformed_anchors)
        classification_loss = torch.randn(batch_size, 4)
        regression_loss = torch.randn(batch_size, 4)
        global_classification_loss = torch.randn(batch_size, 4)
        nms_scores = torch.randn(batch_size, 4)
        global_class = torch.randn(batch_size, 4)
        transformed_anchors = torch.randn(batch_size, 4)
        
        return (classification_loss, regression_loss, global_classification_loss, 
                nms_scores, global_class, transformed_anchors)

# Create a mock dataloader with 3 batches
dataloader_valid = []
for _ in range(3):
    batch = {
        'img': torch.randn(2, 3, 8, 8),
        'annot': torch.randn(2, 5, 4),
        'category': torch.randint(0, 10, (2, 5))
    }
    dataloader_valid.append(batch)

# Create a temporary directory for predictions
predictions_dir = tempfile.mkdtemp()

ARGS = {
    'retinanet': MockRetinaNet(),
    'dataloader_valid': dataloader_valid,
    'epoch_num': 1,
    'predictions_dir': predictions_dir,
    'save_oof': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 3 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.

FunctionAlso present inEarliest-datedOthers named
anchors_for_shape 2 other papers
1 dated earlier, 1 later
1911.09070 (2019-11)
EfficientDet: Scalable and Efficient Object Detection
2203.10981 (2022-03)
MonoDTR: Monocular 3D Object Detection with Depth-Aware Transformer
compute_shape 2 other papers
1 dated earlier, 1 later
1911.09070 (2019-11)
EfficientDet: Scalable and Efficient Object Detection
2203.10981 (2022-03)
MonoDTR: Monocular 3D Object Detection with Depth-Aware Transformer
generate_anchors 2 other papers
1 dated earlier, 1 later
1911.09070 (2019-11)
EfficientDet: Scalable and Efficient Object Detection
2203.10981 (2022-03)
MonoDTR: Monocular 3D Object Detection with Depth-Aware Transformer

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 2026-11-21.

Repositories linked to this paper

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

Abstract

Pneumonia is the leading cause of death among young children and one of the top mortality causes worldwide. The pneumonia detection is usually performed through examine of chest X-ray radiograph by highly-trained specialists. This process is tedious and often leads to a disagreement between radiologists. Computer-aided diagnosis systems showed the potential for improving diagnostic accuracy. In this work, we develop the computational approach for pneumonia regions detection based on single-shot detectors, squeeze-and-excitation deep convolution neural networks, augmentations and multi-task learning. The proposed approach was evaluated in the context of the Radiological Society of North America Pneumonia Detection Challenge, achieving one of the best results in the challenge.

For agents

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

get_harvested_code_for_paper("2005.13899")
get_code_for_paper("2005.13899")
have("2005.13899")

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

curl https://syntology.ai/api/ran/2005.13899.json

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

[![Syntology run record](https://syntology.ai/api/ran/2005.13899.svg)](https://syntology.ai/paper/2005.13899)

Connect an agent — have() is free.