SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1902.11097 · 2019

Predictive Inequity in Object Detection

arXiv · PDF · Open in the Atlas

Code that ran

5 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 0 failures of our instrument; 0 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/1902.11097.json · badge: /api/ran/1902.11097.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
benjaminrwilson/inequity-release community (classifier label)
classifier basis not recorded on the link
5 ran · 0 no recorded run
FunctionStatusWhere it lives
build_ann Ran
ran on a synthesized input
benjaminrwilson/inequity-release/inequity/utils/bdd_to_coco.py
code served (permissive licence) · get_code("1b9ac83ccc95d1c4")
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 dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

BDD_CLASSES = ['bike', 'bus', 'car', 'motor', 'person', 'rider', 'traffic light', 'traffic sign', 'train', 'truck']

torch.manual_seed(0)

# Setup individual arguments
ann = {
    'fitz_category': 'fitz_test',
    'occluded': False,
    'weather': 'sunny',
    'scene': 'street',
    'time_of_day': 'daytime'
}

img_id = '00000'

img_ids = ['00000', '00001', '00002', '00003']

box = [10.0, 20.0, 100.0, 50.0]  # x_min, y_min, width, height

category = 'car'

ann_id = 42

ARGS = {
    'ann': ann,
    'img_id': img_id,
    'img_ids': img_ids,
    'box': box,
    'category': category,
    'ann_id': ann_id
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
coco2bdd Ran
ran on a synthesized input
benjaminrwilson/inequity-release/inequity/utils/utils.py
code served (permissive licence) · get_code("6caf3ae1e1491236")
How it ran

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

torch.manual_seed(0)

# Create a small tensor of COCO labels
# The function checks for labels == 1 (person class in COCO)
# and maps them to 5 (person class in BDD100K)
labels = torch.tensor([0, 1, 2, 1, 3, 4], dtype=torch.long)

ARGS = {
    "labels": labels
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family int, shape (6,) — shape (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
make_coco_annotations Ran
ran on a synthesized input
benjaminrwilson/inequity-release/inequity/utils/bdd_to_coco.py
code served (permissive licence) · get_code("81d8d1977a701a85")
How it ran

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

import json

ARGS = {
    'src_json_file': [
        {
            'category': 'person',
            'bbox': [10, 20, 100, 150],
            'name': 'image_001.jpg',
            'occluded': False,
            'fitz_category': 'type_1',
            'weather': 'clear',
            'scene': 'street',
            'time_of_day': 'daytime'
        },
        {
            'category': 'car',
            'bbox': [50, 60, 200, 180],
            'name': 'image_002.jpg',
            'occluded': True,
            'fitz_category': 'type_2',
            'weather': 'rainy',
            'scene': 'highway',
            'time_of_day': 'night'
        },
        {
            'category': 'rider',
            'bbox': [15, 25, 95, 160],
            'name': 'image_001.jpg',
            'occluded': False,
            'fitz_category': 'type_1',
            'weather': 'clear',
            'scene': 'street',
            'time_of_day': 'daytime'
        }
    ],
    'include': {'type_1', 'type_2'},
    'time_of_day': 'daytime',
    'ignore_occluded': True,
    'file_names': ['image_001.jpg', 'image_002.jpg']
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
make_coco_images Ran
ran on a synthesized input
benjaminrwilson/inequity-release/inequity/utils/bdd_to_coco.py
code served (permissive licence) · get_code("2cb1814334a16263")
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. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import tempfile
import os
from PIL import Image
import numpy as np

# Create a temporary directory and a test image
temp_dir = tempfile.mkdtemp()

# Create a small test image
test_image = Image.new('RGB', (640, 480), color='red')
test_image_path = os.path.join(temp_dir, 'test_image.jpg')
test_image.save(test_image_path)

# Create test JSON data matching BDD100K style
src_json_file = [
    {
        'name': 'test_image',
        'fitz_category': 'type_1',
        'time_of_day': 'day'
    },
    {
        'name': 'test_image',
        'fitz_category': 'type_1',
        'time_of_day': 'night'
    },
    {
        'name': 'test_image_2',
        'fitz_category': 'type_2',
        'time_of_day': 'day'
    },
    {
        'name': 'test_image_3',
        'fitz_category': 'type_3',
        'time_of_day': 'day'
    }
]

# Create another test image for diversity
test_image_2 = Image.new('RGB', (800, 600), color='blue')
test_image_2_path = os.path.join(temp_dir, 'test_image_2.jpg')
test_image_2.save(test_image_2_path)

test_image_3 = Image.new('RGB', (1024, 768), color='green')
test_image_3_path = os.path.join(temp_dir, 'test_image_3.jpg')
test_image_3.save(test_image_3_path)

ARGS = {
    'src_json_file': src_json_file,
    'img_dir': temp_dir,
    'include': {'type_1', 'type_2'},
    'time_of_day': 'day'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, 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
prepare_detections Ran
ran on a synthesized input
benjaminrwilson/inequity-release/inequity/utils/utils.py
code served (permissive licence) · get_code("4d1cefbb6f6fd28c")
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 float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from enum import Enum

class EvalMode(Enum):
    COCO = 0
    FINAL = 1
    ALL = 2

class MockDetections:
    """Mock detections object that supports convert, get_field, and bbox."""
    def __init__(self):
        torch.manual_seed(0)
        self._bbox = torch.tensor([
            [10.0, 20.0, 100.0, 50.0],
            [30.0, 40.0, 80.0, 60.0]
        ])
        self._scores = torch.tensor([0.95, 0.87])
        self._labels = torch.tensor([1.0, 3.0])
        self._converted = False
    
    def convert(self, format_str):
        """Mock convert method that returns self."""
        if format_str == 'xywh':
            self._converted = True
        return self
    
    @property
    def bbox(self):
        """Return bbox tensor."""
        return self._bbox
    
    def get_field(self, field_name):
        """Return field by name."""
        if field_name == 'scores':
            return self._scores
        elif field_name == 'labels':
            return self._labels
        raise ValueError(f"Unknown field: {field_name}")

torch.manual_seed(0)
detections_obj = MockDetections()

ARGS = {
    'img_id': 42,
    'detections': detections_obj,
    'mode': EvalMode.COCO
}
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 list, ndim 2, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 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.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

In this work, we investigate whether state-of-the-art object detection systems have equitable predictive performance on pedestrians with different skin tones. This work is motivated by many recent examples of ML and vision systems displaying higher error rates for certain demographic groups than others. We annotate an existing large scale dataset which contains pedestrians, BDD100K, with Fitzpatrick skin tones in ranges [1-3] or [4-6]. We then provide an in-depth comparative analysis of performance between these two skin tone groupings, finding that neither time of day nor occlusion explain this behavior, suggesting this disparity is not merely the result of pedestrians in the 4-6 range appearing in more difficult scenes for detection. We investigate to what extent time of day, occlusion, and reweighting the supervised loss during training affect this predictive bias.

For agents

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

get_harvested_code_for_paper("1902.11097")
get_code_for_paper("1902.11097")
have("1902.11097")

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

curl https://syntology.ai/api/ran/1902.11097.json

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

[![Syntology run record](https://syntology.ai/api/ran/1902.11097.svg)](https://syntology.ai/paper/1902.11097)

Connect an agent — have() is free.