Intelligent terminal image compression algorithm combining edge calculation and deep learning

Abstract

The invention particularly relates to an intelligent terminal image compression algorithm combining edge calculation and deep learning. The intelligent terminal image compression algorithm combining edge calculation and deep learning is characterized in that an image compression model is deployed in the data transmission process of a terminal and a cloud, image data and video data acquired by an edge terminal are compressed, and key frames and motion information are reserved at the same time, so that data transmission is reduced, bandwidth pressure is reduced, a large amount of storage space is saved, and searching time of positioning abnormal behaviors is shortened. According to the intelligent terminal image compression algorithm combining edge calculation and deep learning, image size of task data with large calculation amount is compressed on the premise of not losing the image precision, the compressed data is transmitted to the cloud for calculation, the edge cooperation advantages of the terminal and the cloud are brought into play to the maximum extent, the network transmission data volume can be reduced, the bandwidth pressure is reduced, the overall response speed is increased, a large amount of storage space can be saved, and the searching time of positioning abnormal behaviors can be shortened.

Description

technical field [0001] The invention relates to the technical fields of edge computing, edge-cloud collaboration, and image compression, and in particular to an image compression algorithm for intelligent terminals that combines edge computing and deep learning. Background technique [0002] Artificial intelligence brings great convenience to social life, but at the same time, there is a great disadvantage. The implementation process requires large-scale data training, excessive reliance on large storage, high-performance hardware devices or cloud resources, forcing it to be used on the terminal Use in smart devices is not widespread. Nowadays, smart terminal devices have gradually become a part of people's lives, and the types of edge devices and the amount of data they generate are increasing sharply. However, the centralized processing mode represented by cloud computing cannot be processed efficiently and timely due to limiting factors such as bandwidth. , unable to mee...

AI Technical Summary

Problems solved by technology

Judging from the current research status, the edge intelligent algorithm can directly use the intelligent algorithm deployed in the device to process the data collected by the device. The advantage is that the data transmission is reduced, and the disadvantage is that the resources such as computing and storage at the edge are insufficient. In order to make better use of edge devices, certain compression operations such as quantization and pruning are usually performed on the model, which can reduce the size of the model to a certain extent, but at the same time it will also lose part of the accuracy. Therefore, when the target image is small or blurred, etc. In image classification applications where recognition is difficult, it is not applicable

Images