一种事件检测模型的自校准优化方法、介质及设备

By collecting scene element data from the target physical location, generating adapted simulation scripts and videos, and optimizing the event detection model, the problem of poor cross-scene adaptability is solved, achieving low-cost, high-efficiency model optimization and accurate detection.

CN121545095BActive Publication Date: 2026-07-17BEIJING QIDAISONG TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIJING QIDAISONG TECH CO LTD
Filing Date
2025-11-10
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing event detection models have poor adaptability across different target physical locations, resulting in high false alarm and false negative rates. Furthermore, they are costly to optimize, lack specificity, and are inefficient.

Method used

By collecting scene element data through a spatial intelligent machine, calling a large text model to generate adapted simulation scripts and videos, and combining the evaluation results to optimize the event detection model, self-calibration is achieved.

Benefits of technology

It reduced data acquisition costs, improved the model's adaptability and optimization efficiency in different scenarios, reduced false positives and false negatives, and enhanced the model's robustness and accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121545095B_ABST
    Figure CN121545095B_ABST
Patent Text Reader

Abstract

本发明涉及模型优化技术领域，尤其涉及一种事件检测模型的自校准优化方法、介质及设备，通过采集多个目标物理场所的场景要素数据，基于场景特征生成适配的模拟剧本与视频，使模型训练数据紧密贴合不同场景的空间布局、环境干扰等特性，结合场景差异化误差进行参数优化，最终目标模型可稳定适配所有目标物理场所，解决了传统模型在跨场景应用中精度骤降的问题；通过文本大模型自动生成正、反例剧本，并经随机化增强生成海量模拟视频，替代了依赖人工拍摄的传统方式，降低了数据获取成本；通过提取中间输出数据，将模型误差量化为阈值过敏感、跟踪丢失、感兴趣区域参数误设等明确类型，使参数优化从经验试错转为定向调参，提高了模型的优化效率。
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Short video intelligent creation system and method based on AI image

    CN119763017A

  • Method and system for quickly generating movie and television animation scene in combination with AI algorithm

    CN120612401A