Localize Any Object in X-Ray Security Scans without Human Annotation

Yaqi Cai; Mingxuan Liu; Lorenzo Vaquero; Ning Wang; Nan Pu; Feng Xue; Elisa Ricci; Nicu Sebe

cs.CVOct 5, 2026

Localize Any Object in X-Ray Security Scans without Human Annotation

Authors: Yaqi Cai, Mingxuan Liu, Lorenzo Vaquero, Ning Wang, Nan Pu, Feng Xue, Elisa Ricci, Nicu Sebe

Organizations: University of Trento · Fondazione Bruno Kessler · Dalian University of Technology · Hefei University of Technology

Abstract

Universal object localization in X-ray security inspection is critical for automated threat detection in safety-critical venues. However, unlike everyday RGB images that dominate web-scale visual data, X-ray scans exhibit distinct color patterns, ambiguous boundaries, and compositional structures caused by volumetric superposition. These gaps hinder the direct zero-shot transfer of dense perception foundation models trained on web-scale RGB data. Moreover, annotated X-ray data is scarce and requires expert labeling, limiting both the training of generalizable X-ray native models and the adaptation of RGB foundation models for X-ray data via fine-tuning. Given these challenges, the bright promise of highly generalizable perception models, enabled by data scaling laws in the RGB domain, remains largely out of reach for X-ray inspection. To this end, we introduce LAO-X, a self-supervised adaptation framework that Locates Any Object in X-ray scans using diverse synthesized image--annotation pairs with granularity-aware supervision. LAO-X first designs a saliency-guided X-ray object mining module to separate diverse object instances, which are then used for physics-guided synthesis in the absorbance domain. LAO-X further incorporates an occlusion-controlled curriculum strategy to fine-tune a Segment Anything Model 2 (SAM2) localizer, progressively adapting it to X-ray scans with increasing object counts and overlap levels. Experiments on six X-ray benchmarks show that LAO-X substantially improves category-agnostic localization, achieving 2% to 23% mAP gains over SAM2 and X-ray specific baselines in heavily cluttered scenarios, entirely without human-annotated labels.

Figures & tables

Figure 1 : Any-object localization in X-ray screening. Top: Point-prompted segmentation on challenging X-ray cases, including hazard objects , unseen objects , common objects with diverse shapes , multi-part instances , and highly transparent items. The green dot indicates the point prompt. Different colors denote predicted masks. Bottom: Radar plots of AP50 , AP75 , and AP ${}_{50}^{95}$ on six benchmarks (Det-Compass, CLCXray, DvXray, PIXray, HiXray, PIDray), comparing SAM2, SAM3, UnSAMv2, and LAO-X (Ours) . LAO-X consistently achieves the strongest performance.

Figure 2 : Overview of LAO-X . Giving an unlabeled X-ray scan, we first extract foreground objects to build an item bank. Then synthesize and annotate images with randomly sampled objects to simulate realistic X-ray overlapping. Finally, we train the model with an occlusion curriculum, which enables multi-granularity localization on X-ray images without manual labels.

Figure 3

Panel of Figure 4 — Figure 4 : Granularity comparison with UnSAMv2. Green overlays are predicted masks at the queried granularity $g$ (shown under each column; same image prompt, only $g$ changes). UnSAMv2 is unstable and may collapse to over-segmenting large regions, while LAO-X follows $g$ more consistently, transitioning from part-level (smaller $g$ ) to holistic masks (larger $g$ ) under heavy overlap.

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 7 : More on comparison of masks with top-5 confidence across real-world X-ray datasets. To highlight the predicted mask, we convert the original image to a grayscale image in our visualizations.

Figure 8 : More on comparison of masks with top-5 confidence across real-world X-ray datasets. To highlight the predicted mask, we convert the original image to a grayscale image in our visualizations.

Evaluation set	$\mathrm{AP}_{50}$
DET-COMPASS, full set	63.39
DET-COMPASS, invisible subset	26.69

Appendix

Table 11 : Localization of low-attenuation objects. We report LAO-X’s box

\mathrm{AP}_{50}

on the full DET-COMPASS evaluation set and its official invisible subset. The lower performance on the invisible subset highlights the difficulty of localizing objects with extremely weak attenuation cues.

Method	$\mathrm{AP}_{50}$	$\mathrm{AP}_{75}$	$\mathrm{AP}_{50:95}$	$\mathrm{AR}_{1}$	$\mathrm{AR}_{10}$
SAM2	0.03	0.02	0.02	0.34	0.36
SAM3	4.18	2.13	2.27	7.74	31.35
UnSAMv2	3.84	1.18	1.61	8.05	19.59
LAO-X	5.87	2.60	2.98	11.92	14.60

Appendix

Table 12 : Localization under extreme overlap on the PIDray Hidden subset. LAO-X achieves the highest box AP and

\mathrm{AR}_{1}

among the evaluated methods, while its lower

\mathrm{AR}_{10}

than SAM3 and UnSAMv2 indicates incomplete recovery of heavily superposed instances. The best results are shown in bold.

Figure 9 : Granularity comparison with UnSAMv2 on CLCXray. Green overlays are predicted masks at the queried granularity $g$ (shown under each column; same image prompt, only $g$ changes). UnSAMv2 is unstable and may collapse to over-segmenting large regions, while LAO-X follows $g$ more consistently, transitioning from part-level (smaller $g$ ) to holistic masks (larger $g$ ) under heavy overlap.

Figure 10 : Granularity sweep of LAO-X. AP 50 as a function of the queried granularity

g\in[0.1,1.0]

on DET-COMPASS and real-world X-ray benchmarks.

Panel of Figure 11 — Figure 11 : More on Granularity comparison with UnSAMv2 on DvXray and HiXray.

Panel of Figure 12 — Figure 12 : More on Granularity comparison with UnSAMv2 on PIXray and PIDray.

Explore similar work

Jun 14, 2026cs.CV

OneFocus: Enabling Real-World X-ray Security Screening with a Unified Vision-Language Model

X-ray contraband detection is critical for security in large-scale logistics and transportation, yet conventional detectors struggle to adapt to emerging contraband types and lack fundamental visual understanding. Vision-language models (VLMs) offer strong generalization but are hindered by the scarcity of high-quality X-ray image-caption data. To bridge this critical gap, we present MMXray, a meticulously curated benchmark of 52,124 image-caption pairs spanning 28 fine-grained classes of X-ray contraband. To enrich MMXray with realistic occlusion patterns, we further introduce CleanDET, a dedicated synthesis dataset containing clean foreground contraband images from 28 categories and background images with diverse density levels, together with AnyContraSyn, a controllable synthesis method designed to operate on CleanDET. We also develop OnePipe, an extensible pipeline for systematic data curation. Built on MMXray, we propose OneFocus, a unified VLM that supports four core tasks: visual question answering, contraband localization, classification, and image understanding. OneFocus achieves state-of-the-art performance in X-ray contraband understanding and demonstrates robust cross-domain generalization, establishing a strong vision-language baseline for security screening.

Jiali Wen, Hongxia Gao, Litao Li +4

2South China University of Technology · 1Xi’an Jiaotong University · 3Shenzhen Loop Area Institute +1

X-Ray Recent Vision-Language Models Screening

Ask PDF

Sep 22, 2026cs.CV

FleXray: Universal Clinical X-ray Segmentation

X-ray is medicine's most widely used imaging modality, yet remains among its least quantitative. Unlike volumetric modalities like CT or MRI, X-ray collapses 3D anatomy into a 2D projection, causing structures to overlap and anatomical boundaries to be ambiguous, even to experts. As a result, labeling X-ray databases for training general-purpose segmentation systems is impractical, leaving morphometric and functional X-ray analysis confined to narrow anatomical regions and applications. To this end, we present FleXray, a generalist model for anatomical segmentation across the entire body in clinical X-rays. Instead of curating large, manually annotated X-ray datasets, we build a scalable, physics-based generative X-ray data engine. Using existing 3D whole-body CT segmentation datasets and generative image-editing models, we simulate fully-annotated 2D X-rays with diverse appearances, physiological properties, and imaging geometries. Trained on these simulations, FleXray accurately segments 60 anatomical structures across unseen research datasets and in-the-wild X-rays. We further show that FleXray makes X-rays directly amenable to quantitative analysis, enabling automated measurements for disease grading, robust navigation during X-ray-guided interventions, and data-efficient learning of pathological targets. We release the model, code, a full-body X-ray segmentation dataset, and a local, easy-to-use browser-based tool at https://flexray.csail.mit.edu .

Victor Ion Butoi, Vivek Gopalakrishnan, John V. Guttag +2

MIT CSAIL · Massachusetts General Hospital · Harvard Medical School

X-Ray Multi Organ Computed Tomography Segmentation X-Ray Photon Correlation Spectroscopy Anatomy

Ask PDF

Jul 15, 2026cs.CV

XCT-SAM: Sequential Parameter-Efficient Domain Adaptation of SAM for Industrial XCT Defect Segmentation

Defect segmentation in additive manufacturing (AM) X-ray computed tomography (XCT) images remains challenging due to severe class imbalance and large distribution shifts across scan conditions. Although recent foundation models such as the Segment Anything Model (SAM) provide strong general-purpose segmentation priors, their natural-image pre-training transfers poorly to the AM XCT domain, where defects appear as subtle non-semantic microstructural anomalies. Moreover, adapting SAM to the AM domain is further limited by the large domain gap and scarcity of labeled real XCT data. We present XCT-SAM, a sequential parameter-efficient adaptation framework for AM XCT defect segmentation. Instead of adapting SAM directly from natural images to XCT data, we first fine-tune Conv-LoRA adapters on an alloy-microstructure dataset and subsequently transfer the adapted model to XCT images, progressively bridging the domain gap. Using Conv-LoRA adapters with rank r=2, the framework injects convolutional spatial inductive bias into SAM's backbone while training approximately 4.15M parameters and keeping over 99% of the model frozen. We evaluate XCT-SAM on out-of-distribution CycleGAN-XCT benchmarks and real-world NIST XCT scans. Across both settings, XCT-SAM consistently outperforms zero-shot SAM and other domain-adapted SAM baselines, achieving the best overall IoU and Dice scores. These results demonstrate the effectiveness of intermediate domain adaptation with parameter-efficient adapters for industrial XCT defect segmentation. The source code is publicly available at https://github.com/Mahedi-61/XCT-SAM.git

Md Mahedi Hasan, Md Mushfiqur Rahaman, Alan Pachkovskiy +3

Lane Department of Computer Science & Electrical Engineering West Virginia University, Morgantown, WV · Department of Industrial and Management Engineering West Virginia University, Morgantown, WV · School of Mathematical and Data Sciences West Virginia University, Morgantown, WV

Defect Detection Domain Adaptation Inductive Bias

Ask PDF