一种面向超高分辨率视频流的智能处理方法
By using parallel inference intelligent acceleration hardware to split, identify, and restore ultra-high resolution video streams, the problem of high-speed identification of ultra-high resolution video streams in existing technologies is solved, and efficient intelligent identification results are achieved.
CN116152059BActive Publication Date: 2026-07-17XIAN AVIATION COMPUTING TECH RES INST OF AVIATION IND CORP OF CHINA
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIAN AVIATION COMPUTING TECH RES INST OF AVIATION IND CORP OF CHINA
- Filing Date
- 2022-12-15
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing image recognition methods and hardware devices are insufficient to support high-speed recognition of ultra-high resolution video streams.
Method used
Parallel inference intelligent acceleration hardware is used, including multiple intelligent computing units and scheduling CPUs. Through decoding, segmentation, parallel processing and splicing, ultra-high resolution video frames are split, identified and restored, and the recognition is accelerated by using embedded neural network processors and CPUs.
Benefits of technology
It achieves high-speed intelligent recognition of ultra-high resolution video streams, is versatile, and adaptable to different application scenarios and target objects.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116152059B_ABST
Abstract
本申请提供了一种面向超高分辨率视频流的智能处理方法，属于智能计算系统技术领域，具体包括通过调度CPU对从视频流信息进行解码还原得到原始高分辨率视频帧信息后进行分割，拆分成多组图像检测算法所需输入尺寸的低分辨率图像，对拆分图像进行编组和保存，推理识别完成后根据编组将处理后的图像覆盖至原图，进行图像还原；将得到的低分辨率图像并行发送至多个智能计算单元中，智能计算单元进行加速识别，主CPU对智能计算单元的识别进行拼接还原为原始图像。通过本申请的处理方案，实现对超高分辨率视频流的高速智能识别。
Need to check novelty before this filing date? Find Prior Art