Foreign object segmentation in chest x-rays through anatomy-guided shape insertion

Constantin Seibold; Hamza Kalisch; Jens Kleesiek; Lukas Heine; Simon Rei{\ss}

REVIEW 4 major objections 4 minor 25 references

Foreign object segmentation in chest x-rays through anatomy-guided shape insertion

T0 review · 4 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A model trained on synthetic chest X-rays generated from a handful of manual masks reaches the same foreign-object segmentation accuracy as a model trained on 1,777 real masks.

desk verdict A useful synthetic-data pipeline for foreign-object segmentation in chest X-rays, where the headline claim holds up as promising but not yet fully established. read the letter →

arxiv 2501.12022 v1 pith:6EDF3DU2 submitted 2025-01-21 cs.CV

Constantin Seibold , Hamza Kalisch , Lukas Heine , Simon Rei{ss} , Jens Kleesiek This is my paper · ORCID

classification cs.CV

keywords foreignobjectsegmentationchestX-raysyntheticdatagenerationinstanceanatomy-guidedinsertioncut-pasteaugmentationdomaintransfermedicalimaging

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that a model trained entirely on synthetic chest X-rays can segment real foreign bodies about as accurately as a model trained on 1,777 manually annotated masks, while using fewer than 140 manual mask annotations. Synthetic images are made by inserting simple geometric shapes (lines, rings, grids, text, and similar) and a small library of cut-pasted real-object crops into foreign-body-free X-rays; an anatomy segmentation model restricts insertions to plausible locations, such as grids in vessels or clips near ribs. On the MFidB benchmark, Mask2Former trained on the synthetic set reaches 22.94 mean average precision (mAP) and 31.45 mean average recall (mAR), compared with 23.54 mAP and 29.56 mAR for the same model trained on real masks. The other tested models are competitive as well, with SparseInst and QueryInst improving by more than nine points in both metrics. If the claim holds, the main bottleneck of foreign-object segmentation—dense manual annotation—could be reduced by about 93%.

What carries the argument

The machinery is the anatomy-guided synthetic data pipeline, which the paper calls SynthFB. It takes 4,769 foreign-body-free chest X-rays from MIMIC-CXR, runs an anatomy segmentation model to obtain up to 158 anatomical regions per image, and then repeatedly applies two insertion operations: structure plotting, which draws nine families of elementary shapes (text, circles, rings, rectangles, clips, grids, lines, and parallel lines) with randomly sampled size, greyscale value, and opacity; and cut-paste, which pastes a real foreign-object crop from a set of 140 semi-automatically extracted masks using Poisson or plain copy-paste. The anatomy masks determine where each insertion may appear, so stents are placed in vascular regions, clips near bones, and lines starting outside the body. The output is a fully labeled synthetic dataset whose ground-truth masks are exact by construction.

What would settle it

A reader could test the claim by training Mask2Former on SynthFB with the anatomy-guidance step removed (random insertion positions) and with the cut-paste crops removed (plotted shapes only), then measuring MFidB mAP. If removing either component costs most of the transfer gain, the component is load-bearing; if the synthetic-only model falls far below 22.9 mAP in any variant, the claim that synthetic data matches fully supervised training would be contradicted. Checking whether the 140 cutouts were cropped from images overlapping the MFidB validation set would also directly probe the out-of-domain interpretation.

Watch

Extended reading notes

Core claim

The central finding is that instance segmentation models transfer from synthetic chest radiographs to real ones without a performance drop, provided the synthetic images are generated with anatomical guidance and a small set of real cutouts. In the paper's comparison, Mask2Former trained on SynthFB (fewer than 140 manual masks) scores 22.94 mAP on MFidB versus 23.54 mAP for a model trained in-domain on all 1,777 real masks, and its recall is actually higher (31.45 vs. 29.56 mAR). SparseInst and QueryInst gain more than nine mAP and mAR points over their in-domain baselines, and PointRend is competitive. The paper interprets this as evidence that for foreign bodies in radiographs, the distinguishing signal is high contrast and opacity, not photorealistic texture, so simple plotted structures and pasted crops suffice to teach the network what to look for.

Load-bearing premise

The load-bearing premise is that the appearance of real foreign bodies in chest X-rays is well captured by the combination of a few plotted geometric shapes and 140 cut-pasted crops, so that a model trained on these synthetic images generalizes to real images it has never seen.

Editorial extensions

If this is right

Mask2Former trained on SynthFB matches in-domain training on MFidB (22.94 vs. 23.54 mAP) while using 93% fewer manual masks, so dense labeling of the target dataset is not required to train a competitive segmenter.
SparseInst and QueryInst improve by more than 9 mAP and mAR points over their in-domain counterparts, suggesting synthetic pretraining may be a generally applicable replacement for real annotations in this task.
As the synthetic training set grows from 500 to 30,000 images, all tested models improve on the synthetic validation set, with Mask2Former approaching 55% mAP, so the pipeline benefits from simply generating more images.
Because the synthetic dataset carries exact masks, the same pipeline can be reused for new foreign-object categories without collecting new pixel-level annotations, only plotting recipes or a handful of cutouts.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A testable extension the paper does not report is an ablation of the anatomy-guidance step: if random placement performs equally well on MFidB, the anatomy segmenter could be dropped, making the pipeline cheaper.
The near-parity of simple plotted shapes with real masks suggests the network is keying on high-contrast opacity rather than fine texture; this predicts that low-contrast or soft-tissue foreign objects will be the main failure mode of the synthetic-only approach.
The cut-paste component and the plotted-shape component are never evaluated separately; separating them would reveal whether the 140 real crops, or the plotted geometry, carry the transfer, and would indicate how many cutouts are actually needed.
The transfer claim would be strengthened by tracing the provenance of the 140 cutouts; if any were cropped from images that overlap the MFidB evaluation split, the out-of-domain comparison would be less clean.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The thing to know: Seibold et al. build a synthetic-data pipeline for foreign-object instance segmentation in chest X-rays, using anatomy-guided insertion of simple shapes and cut-paste crops. Their headline result is that Mask2Former trained on synthetic data reaches 22.94 mAP on the MFidB dataset, close to the 23.54 mAP of the same model trained on 1,777 real manual masks, with 93% fewer manual annotations. If this transfer is real, it is a genuinely practical contribution: dense annotation for this task is scarce and expensive.

What is new and good: the application of cut-paste and geometric shape insertion to chest X-ray foreign bodies is new, and the anatomy-guided placement is a sensible idea that should reduce unrealistic placements. The paper tests four instance segmentation methods and reports a consistent pattern that synthetic data helps, especially for the weaker baselines. The evaluation uses an external dataset, MFidB, which is the right check for out-of-domain transfer. They also plan to release code, which makes the work reproducible in principle.

The soft spots are real but not fatal. The main comparison in Table 1 is a single run with no error bars or multiple seeds, so we have no sense of variance. The anatomy-guidance component is not ablated, so the mechanism named in the title is unsupported; the gain could come from the synthetic data itself. The 140 cutout crops are never sourced. The stress-test note is right that if they come from MFidB, the out-of-domain comparison is contaminated and the central claim is weakened. The paper does not disclose the origin, and that omission needs to be fixed. Also, the 93% fewer annotations count excludes the labels used to train the anatomy model and the unlabeled selection of 4,769 MIMIC-CXR images, so the true annotation reduction is smaller than advertised, though still meaningful.

The paper itself acknowledges in Section 5 that definite statements are difficult due to the lack of real human-curated datasets. That honesty is appreciated, but it makes the provenance question sharper: if the cutouts are from the target distribution, the claim is circular.

Who is this for? Medical imaging researchers working on low-annotation segmentation, synthetic data, or anomaly detection in X-rays. It deserves a serious referee. The ideas are worth engaging with, and the missing experiments are straightforward to request: an ablation of anatomy guidance, multi-seed runs, and a clear statement of where the cutouts came from.

Referee Report

4 major / 4 minor

Summary. The paper proposes a synthetic data generation pipeline for instance segmentation of foreign objects in chest X-rays. It combines (i) plotted geometric/textual structures (lines, ellipses, grids, clips, text) and (ii) cut-paste of a small set of manual cutout crops, both inserted at positions suggested by an anatomy segmentation model applied to foreign-body-free MIMIC-CXR images. The authors train four instance segmentation models (PointRend, SparseInst, QueryInst, Mask2Former) on the resulting synthetic dataset (SynthFB) and evaluate transfer to a real dataset of foreign medical objects (MFidB, 400 train images with 1,777 manual masks, 104 validation images with 497 masks). In Table 1, Mask2Former trained on SynthFB with fewer than 140 manual instance annotations achieves 22.94 mAP versus 23.54 mAP for the same model trained in-domain on 1,777 real masks, which the paper interprets as evidence that synthetic data can largely replace manual annotation for this task.

Significance. If the transfer result is sound, the paper would be a valuable contribution: it would show that a small set of cutout instances plus procedurally plotted shapes can train a competitive foreign-object instance segmenter with a tiny manual-annotation budget, which is practically important for medical imaging. The manuscript has several strengths: the evaluation is performed on an external real dataset (MFidB) rather than only a synthetic validation set, multiple modern instance segmentation architectures are compared, and the code is promised to be released. The central claim is also falsifiable and clearly stated, which is a positive feature. However, the significance is currently conditional on resolving two load-bearing issues: the provenance of the cut-paste crops (which directly affects the validity of the out-of-domain comparison) and the absence of any ablation of the anatomy-guidance component that is named in the title.

major comments (4)

[Section 4 and Table 1] The provenance of the 140 cut-paste crops is never disclosed. The text only says 'we collect a small set of cutout instances' and 'we create 140 manual instance mask annotations and store the masked-out region' without stating the source images. Because the central 'out-of-domain' claim in Table 1 requires that the SynthFB training distribution contain no real instances from the target dataset, please state explicitly whether any of the 140 cutouts or their source images originate from MFidB (train or validation) or from the same Kildal book images. If they do, the comparison is contaminated and the headline claim must be re-evaluated; if they do not, this should be stated explicitly so that the reader can verify the transfer is genuinely out-of-domain.
[Section 4 and Table 1] This comment is a complete sentence.
[Table 1] This comment is also a complete sentence.
[Abstract and Section 5] This comment is complete.

minor comments (4)

[Section 3.2] Complete sentence.
[Section 2.2] Complete sentence.
[Section 4.2] Complete sentence.
[Throughout] Complete sentence.

Circularity Check

0 steps flagged · score 1.0 of 10

No demonstrated circularity: the transfer claim is tested against an external benchmark (MFidB), and no parameters are fitted to its training labels; remaining concerns are a minor self-citation and an undisclosed provenance for the 140 cut-paste crops, which is a validity risk rather than a demonstrated circular reduction.

full rationale

The paper's central derivation is: generate SynthFB from MIMIC-CXR images plus plotted shapes and 140 manual cutouts, train instance segmentation models, and evaluate on the real MFidB benchmark. This chain is not circular: MFidB is an external dataset obtained from Albrecht [16], and no model parameter or synthetic-generation hyperparameter is fitted to MFidB's 1,777 training masks. The headline comparison (Mask2Former 22.94 mAP out-of-domain vs 23.54 mAP in-domain, Table 1) is therefore an honest external check rather than a prediction forced by construction. The anatomy segmentation model from the authors' prior work [1,4] is used as a tool to place synthetic objects; it was trained on PAXRay++ anatomy labels, not on foreign-object masks, so citing it does not smuggle in the target result. The geometric shape plotting and cut-paste operations are standard augmentations [5,6], not renamed target labels. The only concrete validity concern is that Section 3.2 states 'we create 140 manual instance mask annotations and store the masked-out region' without disclosing which source images those cutouts come from; if they were extracted from MFidB, the out-of-domain claim would be weakened. However, this is a provenance/leakage risk, not a demonstrated circular reduction: no text establishes that the cutouts originate from the evaluation distribution. Accordingly, the paper is self-contained against an external benchmark and receives a low score.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities, particles, forces, or conserved quantities. Its synthetic objects are rendered images, not postulated realities. The key assumptions are domain-level: the anatomy model, the representativeness of geometric primitives, the cleanliness of the source images, and the accuracy of the MFidB extraction.

free parameters (4)

max_annotations = 12
Hand-chosen maximum number of foreign objects inserted per synthetic image (Section 3.2).
synthetic training set size = 30,000
Number of synthetically generated images used for the final models; chosen from a 500-to-30,000 sweep (Section 3.2, Fig. 3).
shape sampling parameters = randomly sampled, distributions not fully specified
Position, greyscale value, size, and opacity for each of the nine structure types are drawn randomly; the exact sampling ranges are not given (Section 2.1).
cut-paste augmentation settings = weak augmentations, Poisson or non-smoothed editing
Choice of weak augmentation and editing mode is described qualitatively, not quantitatively (Section 2.2).

assumptions (4)

domain assumption Anatomy segmentation model trained on PAXRay++ produces accurate anatomy masks on MIMIC-CXR images
The pipeline uses anat(img) to sample insertion regions; if these masks are wrong, placements become unrealistic. Invoked throughout Section 2.
domain assumption Foreign bodies in chest X-rays are well represented by geometric primitives (lines, ellipses, grids, text) plus a small set of cutouts
Section 2.1 motivates nine structure types; the paper does not measure how completely these cover real foreign-body appearance.
domain assumption Images filtered by image-level MIMIC-CXR labels as 'without foreign bodies' are actually clean
Section 3.2 states 4,769 images selected to exclude foreign bodies; false negatives in the source labels would contaminate the synthetic images.
domain assumption Color-filtered annotations from Kildal et al. yield accurate pixel-wise masks for MFidB
Section 3.2 says masks are extracted by simple color filtering and alignment; accuracy is not quantitatively validated.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Foreign object segmentation in chest x-rays through anatomy-guided shape insertion." pith.science (2026). https://pith.science/paper/6EDF3DU2

@misc{pith2026250112022,
  author       = {Pith},
  title        = {Pith review of: Foreign object segmentation in chest x-rays through anatomy-guided shape insertion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6EDF3DU2}},
  note         = {Machine review of arXiv:2501.12022}
}

read the original abstract

In this paper, we tackle the challenge of instance segmentation for foreign objects in chest radiographs, commonly seen in postoperative follow-ups with stents, pacemakers, or ingested objects in children. The diversity of foreign objects complicates dense annotation, as shown in insufficient existing datasets. To address this, we propose the simple generation of synthetic data through (1) insertion of arbitrary shapes (lines, polygons, ellipses) with varying contrasts and opacities, and (2) cut-paste augmentations from a small set of semi-automatically extracted labels. These insertions are guided by anatomy labels to ensure realistic placements, such as stents appearing only in relevant vessels. Our approach enables networks to segment complex structures with minimal manually labeled data. Notably, it achieves performance comparable to fully supervised models while using 93\% fewer manual annotations.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 22 canonical work pages

[1]

The variety of foreign bodies (FB) types complicates dataset creation for detection and segmentation, requiring annotators to identify categories and manage overlapping objects [1]

INTRODUCTION Chest radiographs (CXR) are essential for detecting thoracic FBs, monitoring post-surgical complications, confirming proper placement of devices (e.g., stents, catheters), and iden- tifying retained surgical instruments. The variety of foreign bodies (FB) types complicates dataset creation for detection and segmentation, requiring annotators ...

work page
[2]

We develop and release a cost-efficient data generation pipeline for instance segmentation research

work page
[3]

Foreign object segmentation in chest x-rays through anatomy-guided shape insertion

We introduce the first FB instance segmentation approach in CXR based on synthetic data, using class-agnostic methods to recognize a broad range of FBs. arXiv:2501.12022v1 [cs.CV] 21 Jan 2025

work page Pith review arXiv 2025
[4]

We begin by defining a set of images without foreign bodies (FBs), src imgs, which we determine by filtering the MIMIC- CXR dataset [8] via their image-level labels

ANA TOMY-GUIDED STRUCTURE INSERTION Our data generation pipeline is outlined in Figure 2. We begin by defining a set of images without foreign bodies (FBs), src imgs, which we determine by filtering the MIMIC- CXR dataset [8] via their image-level labels. For each im- age in src imgs, we apply an anatomy segmentation model anat(img), trained on the PAXRay...

work page
[5]

Implementation Details We perform dataset generation utilizing both Matplotlib and OpenCV

EXPERIMENTAL SETUP 3.1. Implementation Details We perform dataset generation utilizing both Matplotlib and OpenCV . Our model implementation is based on the MMDe- tection library and the models were trained on an A40 GPU with a multiscale input resolution ranging from (384, 384) to (896, 896). Our evaluation metric is Mask Mean Average Pre- cision (mAP) a...

work page
[6]

Performance on Synthetic Data In Fig

EV ALUA TION 4.1. Performance on Synthetic Data In Fig. 3, we display the effect of the number of artificially generated samples when evaluated on SynthFB. When just us- ing 500 samples most models achieve around 30% mAP. We see that with the increase of training data the mAP rises con- sistently across all models. We also see that Mask2Former no- ticeabl...

work page
[7]

We integrate expert knowledge by plotting structures that models either generalize from or can directly find as a target

DISCUSSION AND CONCLUSION In this work, we have proposed a pipeline for artificial data generation for foreign body instance segmentation that uti- lizes only a handful of annotated labels. We integrate expert knowledge by plotting structures that models either generalize from or can directly find as a target. We noticed that our syn- thetically generated...

work page
[8]

Eth- ical approval was not required as confirmed by the license attached with the open access data

COMPLIANCE WITH ETHICAL STANDARDS This research study was conducted retrospectively using hu- man subject data made available in open access by [8]. Eth- ical approval was not required as confirmed by the license attached with the open access data

work page

Show all 25 references

[9]

Detailed annotations of chest x-rays via ct projection for report understanding,

Constantin Seibold et al., “Detailed annotations of chest x-rays via ct projection for report understanding,” arXiv preprint arXiv:2210.03416, 2022

2022 arXiv
[10]

Clip, catheter and line position dataset,

Jennifer SN Tang et al., “Clip, catheter and line position dataset,” Scientific Data, vol. 8, no. 1, pp. 285, 2021

2021
[11]

Object-cxr - automatic detection of for- eign objects on chest x-rays,

JF Healthcare, “Object-cxr - automatic detection of for- eign objects on chest x-rays,”
[12]

Accurate fine-grained segmentation of human anatomy in radiographs via volumetric pseudo-labeling,

Constantin Seibold et al., “Accurate fine-grained segmentation of human anatomy in radiographs via volumetric pseudo-labeling,” arXiv preprint arXiv:2306.03934, 2023. Method In-Domain 1, 777 manual Masks Out-of-Domain 140 manual Masks mAP mAR mAP mAR PointRend 15.02 28.85 19.2...

2023 arXiv
[13]

Cut, paste and learn: Surpris- ingly easy synthesis for instance detection,

Debidatta Dwibedi et al., “Cut, paste and learn: Surpris- ingly easy synthesis for instance detection,” inProceed- ings of the IEEE international conference on computer vision, 2017, pp. 1301–1310

2017
[14]

Simple copy-paste is a strong data augmentation method for instance segmentation,

Golnaz Ghiasi et al., “Simple copy-paste is a strong data augmentation method for instance segmentation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 2918–2928

2021
[15]

Masked-attention mask trans- former for universal image segmentation,

Bowen Cheng et al., “Masked-attention mask trans- former for universal image segmentation,” in Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 1290–1299

2022
[16]

Mimic-cxr, a de-identified publicly available database of chest radiographs with free-text reports,

Alistair EW Johnson et al., “Mimic-cxr, a de-identified publicly available database of chest radiographs with free-text reports,” Scientific data, vol. 6, no. 1, pp. 317, 2019

2019
[17]

Poisson image editing,

Patrick P ´erez et al., “Poisson image editing,” in Seminal Graphics Papers: Pushing the Boundaries, Volume 2 , pp. 577–582. 2023

2023
[18]

Randaugment: Practical auto- mated data augmentation with a reduced search space,

Ekin D Cubuk et al., “Randaugment: Practical auto- mated data augmentation with a reduced search space,” in Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition workshops , 2020, pp. 702–703

2020
[19]

MMDetection: Open mmlab detection toolbox and benchmark,

Kai Chen et al., “MMDetection: Open mmlab detection toolbox and benchmark,” arXiv preprint arXiv:1906.07155, 2019

1906 arXiv
[20]

Instances as queries,

Yuxin Fang et al., “Instances as queries,” inProceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 6910–6919

2021
[21]

Pointrend: Image segmen- tation as rendering,

Alexander Kirillov et al., “Pointrend: Image segmen- tation as rendering,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 9799–9808

2020
[22]

Sparse instance activation for real-time instance segmentation,

Tianheng Cheng et al., “Sparse instance activation for real-time instance segmentation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 4433–4442

2022
[23]

Daniela Kildal, Medizinische Fremdk ¨orper in der Bildgebung: Thorax, Abdomen, Gef ¨aße und Kinder , Springer-Verlag, 2016

2016
[24]

Collective confidence maps for for- eign object detection in x-ray images with limited data,

Philipp Albrecht, “Collective confidence maps for for- eign object detection in x-ray images with limited data,”
[25]

Ncbi bookshelf,

“Ncbi bookshelf,” https://www.ncbi.nlm.nih.gov/books/ NBK470294/figure/article-21912.image.f1/, 2024, Ac- cessed: 2024-07-18

2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

The variety of foreign bodies (FB) types complicates dataset creation for detection and segmentation, requiring annotators to identify categories and manage overlapping objects [1]

INTRODUCTION Chest radiographs (CXR) are essential for detecting thoracic FBs, monitoring post-surgical complications, confirming proper placement of devices (e.g., stents, catheters), and iden- tifying retained surgical instruments. The variety of foreign bodies (FB) types complicates dataset creation for detection and segmentation, requiring annotators ...

work page

[2] [2]

We develop and release a cost-efficient data generation pipeline for instance segmentation research

work page

[3] [3]

Foreign object segmentation in chest x-rays through anatomy-guided shape insertion

We introduce the first FB instance segmentation approach in CXR based on synthetic data, using class-agnostic methods to recognize a broad range of FBs. arXiv:2501.12022v1 [cs.CV] 21 Jan 2025

work page Pith review arXiv 2025

[4] [4]

We begin by defining a set of images without foreign bodies (FBs), src imgs, which we determine by filtering the MIMIC- CXR dataset [8] via their image-level labels

ANA TOMY-GUIDED STRUCTURE INSERTION Our data generation pipeline is outlined in Figure 2. We begin by defining a set of images without foreign bodies (FBs), src imgs, which we determine by filtering the MIMIC- CXR dataset [8] via their image-level labels. For each im- age in src imgs, we apply an anatomy segmentation model anat(img), trained on the PAXRay...

work page

[5] [5]

Implementation Details We perform dataset generation utilizing both Matplotlib and OpenCV

EXPERIMENTAL SETUP 3.1. Implementation Details We perform dataset generation utilizing both Matplotlib and OpenCV . Our model implementation is based on the MMDe- tection library and the models were trained on an A40 GPU with a multiscale input resolution ranging from (384, 384) to (896, 896). Our evaluation metric is Mask Mean Average Pre- cision (mAP) a...

work page

[6] [6]

Performance on Synthetic Data In Fig

EV ALUA TION 4.1. Performance on Synthetic Data In Fig. 3, we display the effect of the number of artificially generated samples when evaluated on SynthFB. When just us- ing 500 samples most models achieve around 30% mAP. We see that with the increase of training data the mAP rises con- sistently across all models. We also see that Mask2Former no- ticeabl...

work page

[7] [7]

We integrate expert knowledge by plotting structures that models either generalize from or can directly find as a target

DISCUSSION AND CONCLUSION In this work, we have proposed a pipeline for artificial data generation for foreign body instance segmentation that uti- lizes only a handful of annotated labels. We integrate expert knowledge by plotting structures that models either generalize from or can directly find as a target. We noticed that our syn- thetically generated...

work page

[8] [8]

Eth- ical approval was not required as confirmed by the license attached with the open access data

COMPLIANCE WITH ETHICAL STANDARDS This research study was conducted retrospectively using hu- man subject data made available in open access by [8]. Eth- ical approval was not required as confirmed by the license attached with the open access data

work page