Computer-implemented method for detecting one or more defined objects from a plurality of single-object images obtained from a multi-object image, such as a scene image
The ELM-based object detection method addresses resource and scalability challenges in multi-object images by optimizing training and inference processes, leveraging quantum computing for efficient and accurate object detection.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- QUANTUMBASEL AG
- Filing Date
- 2025-12-12
- Publication Date
- 2026-07-02
AI Technical Summary
Existing object detection methods in multi-object images face challenges with high computational resource demands, time-consuming annotation processes, and lack of interpretability, limiting their scalability and accessibility, especially in resource-constrained environments.
A computer-implemented method using Extreme Learning Machines (ELM) with an attention-based Multiple Instance Learning model for object detection from single-object images, optimized for reduced computational and data resource requirements, leveraging quantum computing for enhanced efficiency and accuracy.
The method achieves faster training and inference, improved accuracy, and reduced energy consumption, enabling real-time decision-making and robust handling of noisy data, while maintaining model interpretability.
Smart Images
Figure EP2025086785_02072026_PF_FP_ABST