RT Conference_Proceedings
T1 Superpowering Open-Vocabulary Object Detectors for X-ray Vision
A1 García Fernández, Pablo
A1 Vaquero Otal, Lorenzo
A1 Liu, Mingxuan
A1 Xue, Feng
A1 Cores Costa, Daniel
A1 Sebe, Nicu
A1 Mucientes Molina, Manuel
A1 Ricci, Elisa
K1 X-ray
K1 Open-vocabulary object detection
AB Open-vocabulary object detection (OvOD) is set to revolutionize security screening by enabling systems to recognize any item in X-ray scans. However, developing effective OvOD models for X-ray imaging presents unique challenges due to data scarcity and the modality gap that prevents direct adoption of RGB-based solutions. To overcome these limitations, we propose RAXO, a training-free framework that repurposes off-the-shelf RGB OvOD detectors for robust X-ray detection. RAXO builds high-quality X-ray class descriptors using a dual-source retrieval strategy. It gathers relevant RGB images from the web and enriches them via a novel X-ray material transfer mechanism, eliminating the need for labeled databases. These visual descriptors replace text-based classification in OvOD, leveraging intra-modal feature distances for robust detection. Extensive experiments demonstrate that RAXO consistently improves OvOD performance, providing an average mAP increase of up to 17.0 points over base detectors. To further support research in this emerging field, we also introduce DET-COMPASS, a new benchmark featuring bounding box annotations for over 300 object categories, enabling large-scale evaluation of OvOD in X-ray.
YR 2025
FD 2025-10-19
LK https://hdl.handle.net/10347/43850
UL https://hdl.handle.net/10347/43850
LA eng
NO International Conference on Computer Vision, ICCV 2025, Honolulu 2025
NO We thank CINECA and the ISCRA initiative for the availability of high-performance computing resources. This work was partially supported by the EU HORIZON IAMI (HORIZON-CL3-2023-FCT-01-04-101168272) project, the EU HORIZON ELIAS (HORIZONCL4-2022-HUMAN-02-101120237) project, the EU ISFP PRECRISIS (ISFP-2022-TFI-AG-PROTECT-02101100539) project, the MUR PNRR FAIR (PE00000013) project funded by the NextGenerationEU, the Spanish Ministerio de Ciencia e Innovación (grant numbers PID2020-112623GB-I00, PID2023-149549NB-I00), and the Galician Consellería de Cultura, Educación e Universidade (2024-2027 ED431G-2023/04). Some of these grants are co-funded by the European Regional Development Fund (ERDF). Pablo Garcia-Fernandez is supported by the Spanish Ministerio de Universidades under the FPU national plan (grant number FPU21/05581).
DS Minerva
RD 24 abr 2026