一种目标识别方法、装置、服务器及存储介质
By predicting the location of targets in the same row and avoiding duplicate recognition areas in target recognition, the problem of resource waste caused by multiple similar images is solved, and more efficient target recognition is achieved.
CN117809159BActive Publication Date: 2026-07-17ZHEJIANG UNIVIEW TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHEJIANG UNIVIEW TECH CO LTD
- Filing Date
- 2023-08-09
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In existing technologies, receiving multiple similar scene images during target recognition can lead to duplicate targets being identified, increasing computational power consumption and wasting resources.
Method used
By acquiring scene images from the same acquisition device at two different acquisition times, the set of targets in the same row and their relative position information are determined, the position of the target at the next acquisition time is predicted, and the position area of the already identified target is designated as the non-detection area to reduce duplicate identification.
Benefits of technology
It effectively avoids repeated target identification, reduces computing power consumption and resource waste, and improves identification efficiency.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN117809159B_ABST
Abstract
本发明公开了一种目标识别方法、装置、服务器及存储介质。该方法包括：获取第一场景图像和第二场景图像；其中，第一场景图像中包含至少两个检测目标；从至少两个检测目标中确定存在同行关系的目标集合，并在第一场景图像中确定目标集合中各个第一目标间的相对位置信息；确定目标时间间隔后第二目标在第二场景图像中的第一预测位置；其中，第二目标为目标集合中的任一第一目标；根据第一预测位置及相对位置信息确定第三目标在第二场景图像中所处的第二预测位置；其中，第三目标为目标集合中除第二目标外的第一目标；对第二场景图像中除第一预测位置及第二预测位置外的区域进行目标识别。本方案解决了识别出的目标存在重复造成资源浪费的问题。
Need to check novelty before this filing date? Find Prior Art