SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.06364 · ICML · 2020

On uncertainty estimation in active learning for image segmentation

Bo Li, Tommy Alstrøm

arXiv · PDF · Open in the Atlas

Code that ran

10 of 18 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 9 have no contract verdict; 0 failures of our instrument; 8 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/2007.06364.json · badge: /api/ran/2007.06364.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
lyn1874/region_based_active_learning canonical (classifier label)
classifier basis: census
10 ran · 8 no recorded run
FunctionStatusWhere it lives
ax_global_get Ran
ran on a synthesized input
lyn1874/region_based_active_learning/visualize_calibration_score.py
code served (permissive licence) · get_code("50fc13a68b64379f")
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 Axes. 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 matplotlib.pyplot as plt

fig = plt.figure()

ARGS = {
    'fig': fig
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Axes — a Axes output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Axes 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
calc_uncertainty Ran
ran on a synthesized input
lyn1874/region_based_active_learning/eval_calibration/calc_calibration_score.py
code served (permissive licence) · get_code("029462f1307151a0")
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 probability arrays for different methods
# Method B and C: simple probability distribution
prob_bc = np.array([[0.7, 0.2, 0.05, 0.05],
                     [0.1, 0.3, 0.4, 0.2]])

# Method D: tuple of (prob, bald)
# prob shape: (2, 4) - batch of 2, 4 classes
# bald shape: (2, 4) - batch of 2, 4 classes (uncertainty per class)
prob_d_main = np.array([[0.7, 0.2, 0.05, 0.05],
                        [0.1, 0.3, 0.4, 0.2]])
prob_d_bald = np.array([[0.05, 0.03, 0.02, 0.01],
                        [0.04, 0.06, 0.03, 0.02]])
prob_d = (prob_d_main, prob_d_bald)

ARGS = {
    'prob': prob_bc,
    'method': 'B',
    'reshape': True
}
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
calculate_score_for_patch Ran
ran on a synthesized input
lyn1874/region_based_active_learning/models/acquisition_region.py
code served (permissive licence) · get_code("a767dc35a3f40491")
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 numpy as np
from scipy import signal, ndimage

np.random.seed(0)

# Create small test uncertainty estimates: [num_images, height, width]
uncert_est = np.random.rand(2, 8, 8).astype(np.float32)

# Create a small kernel for patch selection
kernel = np.ones((3, 3)).astype(np.float32)

# Stride size for patch sampling
stride_size = 1

# Number of most uncertain patches to select
num_most_uncertain_patch = 4

ARGS = {
    "uncert_est": uncert_est,
    "kernel": kernel,
    "stride_size": stride_size,
    "num_most_uncertain_patch": num_most_uncertain_patch,
}
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 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
extract_bald_index Ran
ran on a synthesized input
lyn1874/region_based_active_learning/models/acquistion_full_image.py
code served (permissive licence) · get_code("e140b3c7b79e9645")
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, 1) 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 test arrays matching the function's expected shapes
# fb_prob_mean_bald: shape [Number_of_Image, im_h, im_w, 2]
# fb_prob: shape [Number_of_Image, im_h, im_w, 2]
# x_image_pl: shape [num_image, imh, imw, 3]

num_images = 2
im_h = 4
im_w = 5
num_channels_image = 3
num_classes = 2

# Create fb_prob_mean_bald with small positive values
# This represents 1/t*p_c*log(p_c) values
fb_prob_mean_bald = np.random.rand(num_images, im_h, im_w, num_classes) * 0.1

# Create fb_prob with valid probability values (sum to ~1 along last axis)
fb_prob_raw = np.random.rand(num_images, im_h, im_w, num_classes)
fb_prob = fb_prob_raw / (np.sum(fb_prob_raw, axis=-1, keepdims=True) + 1e-8)

# Create x_image_pl with some non-zero regions
# Using distinct height (4) and width (5) to avoid ambiguity
x_image_pl = np.random.rand(num_images, im_h, im_w, num_channels_image) * 255
# Set some values to 0 to test the sele_index logic
x_image_pl[0, 0, :, :] = 0
x_image_pl[1, 3, :, :] = 0

# Test both aggregation methods
agg_method_sum = "Simple_Sum"
agg_method_quantile = "Quantile"
quantile_cri = 75  # 75th percentile

ARGS = {
    "fb_prob_mean_bald": fb_prob_mean_bald,
    "fb_prob": fb_prob,
    "x_image_pl": x_image_pl,
    "agg_method": agg_method_sum,
    "quantile_cri": quantile_cri
}
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, 1) — shape (2, 1) 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_entropy_index Ran
ran on a synthesized input
lyn1874/region_based_active_learning/models/acquistion_full_image.py
code served (permissive licence) · get_code("87aacc015d138bc6")
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, 1) 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 test data
np.random.seed(0)

# Create fb_prob: shape (batch, height, width, channels)
# Using shape (2, 4, 5, 3) for distinct dimensions
fb_prob = np.random.dirichlet(np.ones(3), size=(2, 4, 5)).astype(np.float32)
fb_prob = np.transpose(fb_prob, (0, 1, 2, 3))  # Shape: (2, 4, 5, 3)

# Create images: shape (batch, height, width, channels)
# Most values non-zero, with some zero regions to test the masking logic
images = np.ones((2, 4, 5, 3), dtype=np.float32)
images[0, 0:2, 0:2, :] = 0  # Zero region in first image
images[1, 3, 4, :] = 0      # Small zero region in second image

# Set aggregation method
agg_method = 'Quantile'

# Set quantile criterion
quantile_cri = 75

ARGS = {
    'fb_prob': fb_prob,
    'images': images,
    'agg_method': agg_method,
    'quantile_cri': quantile_cri
}
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, 1) — shape (2, 1) 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_uncertainty_index Ran
ran on a synthesized input
lyn1874/region_based_active_learning/models/acquistion_full_image.py
code served (permissive licence) · get_code("8c7ef5678477ae92")
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, 1) 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 data
# images: shape (batch_size, height, width, channels)
# The function uses np.mean(images[i, :, :, :], -1) to get a 2D mean across channels
# It checks where this mean != 0 to find non-zero regions
batch_size = 2
height = 5
width = 6
channels = 3

# Create images with some non-zero regions
images = np.random.rand(batch_size, height, width, channels) * 255
# Ensure some regions are zero for realistic data
images[0, 0:2, 0:2, :] = 0

# fb_prob: shape (batch_size, height, width, 2)
# The function accesses fb_prob[i, :, :, 0] and fb_prob[i, :, :, 1]
# These represent probabilities for 2 classes
fb_prob = np.random.rand(batch_size, height, width, 2)
# Normalize to be valid probabilities
fb_prob = fb_prob / np.sum(fb_prob, axis=-1, keepdims=True)

# agg_method: string parameter, either 'Simple_Sum' or 'Quantile'
agg_method = 'Quantile'

# quantile_cri: quantile criterion (0-100 scale for np.percentile)
quantile_cri = 75

ARGS = {
    'images': images,
    'fb_prob': fb_prob,
    'agg_method': agg_method,
    'quantile_cri': quantile_cri
}
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, 1) — shape (2, 1) 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
get_time Ran
ran on a synthesized input
lyn1874/region_based_active_learning/eval_calibration/calc_calibration_score.py
code served (permissive licence) · get_code("1ea8d82f1e501df8")
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 time

# The function takes time_init (a float representing initial time) 
# and opt (a string for the option/label to print)

time_init = time.time()
opt = "model_training"

ARGS = {
    "time_init": time_init,
    "opt": opt
}
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 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
give_score_path Ran
ran on a synthesized input
lyn1874/region_based_active_learning/visualize_calibration_score.py
code served (permissive licence) · get_code("486f99c1fc0d5a72")
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 3, 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 os
import tempfile
import shutil

# Create a temporary directory structure for testing
temp_base = tempfile.mkdtemp()

# Create region_calibration_stat directory with .obj files
region_dir = os.path.join(temp_base, 'region_calibration_stat')
os.makedirs(region_dir, exist_ok=True)

# Create sample .obj files with region markers
region_files = [
    'model_B_001.obj',
    'model_B_002.obj',
    'model_C_001.obj',
    'model_C_002.obj',
    'model_D_001.obj',
    'model_D_002.obj',
]

for filename in region_files:
    filepath = os.path.join(region_dir, filename)
    with open(filepath, 'w') as f:
        f.write('# dummy obj file\n')

# Create full_image_calibration_stat directory with .obj files
full_dir = os.path.join(temp_base, 'full_image_calibration_stat')
os.makedirs(full_dir, exist_ok=True)

full_files = [
    'full_B_001.obj',
    'full_B_002.obj',
    'full_C_001.obj',
    'full_C_002.obj',
    'full_D_001.obj',
    'full_D_002.obj',
]

for filename in full_files:
    filepath = os.path.join(full_dir, filename)
    with open(filepath, 'w') as f:
        f.write('# dummy obj file\n')

# Define path with trailing slash
path = temp_base + '/'

ARGS = {
    'path': path
}
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 tuple, ndim 3, 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
str2bool Ran
ran, violated the contract we drafted
lyn1874/region_based_active_learning/visualize_calibration_score.py
code served (permissive licence) · get_code("7c508037b40522af")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned 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 argparse

def str2bool(v):
    if isinstance(v, bool):
        return v
    if v.lower() in ('yes', 'true', 't', 'y', '1'):
        return True
    elif v.lower() in ('no', 'false', 'f', 'n', '0'):
        return False
    else:
        raise argparse.ArgumentTypeError('Boolean value expected.')

# Test cases for str2bool
ARGS = {
    'v': 'true'  # string input that should convert to boolean
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 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
transfer_numeric_index_back_to_imindex Ran
ran on a synthesized input
lyn1874/region_based_active_learning/eval_calibration/calc_calibration_score.py
code served (permissive licence) · get_code("16bcb3740ed61a09")
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 int, shape (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

# Construct a valid numeric_index array
# The function expects an array where each element is an index into im_index_pool
# Since im_index_pool has 65 elements (0-64), numeric_index should contain valid indices
# We'll create a small array with valid indices (less than 65)
numeric_index = np.array([5, 10, 2, 30], dtype='int64')

ARGS = {
    'numeric_index': numeric_index
}
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 int, shape (4,) — shape (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
calc_auc_score Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/optimization/loss_region_specific.py
code served (permissive licence) · get_code("d2b70422b98f5661")
How it ran

no invocation recorded.

calc_f1_score Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/optimization/loss_region_specific.py
code served (permissive licence) · get_code("78f559a5e9a51a8b")
How it ran

no invocation recorded.

conv_layer_renew Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/models/layer_utils.py
code served (permissive licence) · get_code("a3a2aaecc255402f")
How it ran

no invocation recorded.

deconv_layer_renew Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/models/layer_utils.py
code served (permissive licence) · get_code("12d68c73441b00a9")
How it ran

no invocation recorded.

get_deconv_layer_weight Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/models/layer_utils.py
code served (permissive licence) · get_code("1229928c1c5f8499")
How it ran

no invocation recorded.

return_pseudo_label Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/models/acquisition_region.py
code served (permissive licence) · get_code("05677c6af8f1ea2d")
How it ran

no invocation recorded.

select_most_uncertain_patch Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/models/acquisition_region.py
code served (permissive licence) · get_code("c1fe50bec09abed1")
How it ran

no invocation recorded.

tf_f1score Not yet run
unverified (harvested, no recorded run)
lyn1874/region_based_active_learning/optimization/loss_region_specific.py
code served (permissive licence) · get_code("95bd0e68182d1059")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 1 later) and 17 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
str2bool 175 other papers
19 dated earlier, 1 the same month, 155 later · 2 only as an assembly brick
1406.6247 (2014-06)
Recurrent Models of Visual Attention
1603.09382 (2016-03)
Deep Networks with Stochastic Depth
1710.10196 (2017-10)
Progressive Growing of GANs for Improved Quality, Stability, and Variation
and 172 more not named here

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

Uncertainty estimation is important for interpreting the trustworthiness of machine learning models in many applications. This is especially critical in the data-driven active learning setting where the goal is to achieve a certain accuracy with minimum labeling effort. In such settings, the model learns to select the most informative unlabeled samples for annotation based on its estimated uncertainty. The highly uncertain predictions are assumed to be more informative for improving model performance. In this paper, we explore uncertainty calibration within an active learning framework for medical image segmentation, an area where labels often are scarce. Various uncertainty estimation methods and acquisition strategies (regions and full images) are investigated. We observe that selecting regions to annotate instead of full images leads to more well-calibrated models. Additionally, we experimentally show that annotating regions can cut 50% of pixels that need to be labeled by humans compared to annotating full images.

For agents

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

get_harvested_code_for_paper("2007.06364")
get_code_for_paper("2007.06364")
have("2007.06364")

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

curl https://syntology.ai/api/ran/2007.06364.json

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

[![Syntology run record](https://syntology.ai/api/ran/2007.06364.svg)](https://syntology.ai/paper/2007.06364)

Connect an agent — have() is free.