EUVIMEDEUVIMED entdecken
Crossref

Synthetic Priors for Real-World Detection: A Label-Free Framework for Identifying Ultra-Rare Objects

Di Wang, Tongning Wu, Lei Yang, Xianjun Fan, Xin Ye, Renbing Huang, Xing Lu, Xu Xu, Congsheng Li

IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026

DOI: 10.1109/tpami.2026.3693305 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026. Verfasst von Di Wang und 8 weiteren Autor:innen. Hauptthemen laut MEDLINE: Pattern Recognition, Automated, Image Processing, Computer-Assisted. Die Arbeit nennt 40 Literaturangaben. Gefördert von: Zhuhai City Industry-University-Research Cooperation, Noncommunicable Chronic Diseases-National Science and Technology Major.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Detecting ultra-rare objects (frequency < 10-6) without positive annotations remains a critical challenge in computer vision. We propose a Synthesis-Prior Driven paradigm that eliminates the need for real positive samples. Our core insight is that models trained on statistically and perceptually indistinguishable synthetic data can generalize effectively to real-world environments. To achieve this, we introduce a Differentiable Morphological Modeling framework, which embeds explicit physical and geometric priors into a differentiable synthesis engine. Unlike deep learning generative models, this approach ensures high-fidelity imagery with perfectly accurate, objective annotations. We validated this paradigm on the safety-critical task of detecting Circulating Genetically Abnormal Cells (CACs). Perceptual assessments confirmed that synthetic data is indistinguishable from clinical imagery (expert accuracy ≈ 50%). Consequently, our model, trained exclusively on synthetic data, achieved state-of-the-art performance on real patient samples (98.43% accuracy, 97.40% sensitivity), outperforming both human experts and existing automated methods. Furthermore, the model demonstrated exceptional robustness across a multi-center cohort (CV < 0.8%) and achieved high-precision zero-shot transfer in industrial defect inspection (DAGM 2007). This study establishes a principled methodology for ultra-rare object detection, offering broad applicability across domains with definable geometric priors.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Pattern Recognition, AutomatedImage Processing, Computer-AssistedHumansDetection AlgorithmsAlgorithmsGenerative Artificial IntelligenceDeep Learning

Förderung

  • Zhuhai City Industry-University-Research Cooperation
  • Noncommunicable Chronic Diseases-National Science and Technology Major

Zitieren

APA

Wang, D., Wu, T., Yang, L., Fan, X., Ye, X., Huang, R., Lu, X., Xu, X., & Li, C. (2026). Synthetic Priors for Real-World Detection: A Label-Free Framework for Identifying Ultra-Rare Objects. IEEE Transactions on Pattern Analysis and Machine Intelligence. https://doi.org/10.1109/tpami.2026.3693305

Vancouver

Wang D, Wu T, Yang L, Fan X, Ye X, Huang R, et al. Synthetic Priors for Real-World Detection: A Label-Free Framework for Identifying Ultra-Rare Objects. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026. doi:10.1109/tpami.2026.3693305

BibTeX
@article{Wang2026,
  author = {Wang, Di and Wu, Tongning and Yang, Lei and Fan, Xianjun and Ye, Xin and Huang, Renbing and Lu, Xing and Xu, Xu and Li, Congsheng},
  title = {Synthetic Priors for Real-World Detection: A Label-Free Framework for Identifying Ultra-Rare Objects},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2026},
  doi = {10.1109/tpami.2026.3693305}
}
Kostenlos starten →