An image sample data set construction method and device based on dynamic panoramic image reconstruction and terminal equipment

Abstract

The invention discloses an image sample data set construction method and device based on dynamic panoramic image reconstruction and terminal equipment. Comprising the steps of performing video recording on a sampling scene by using a camera group; performing frame synchronization on the plurality of recorded videos to form a frame group; calculating the mutual relation among the plurality of camera images at the same time, and laminating the camera images into a panoramic image of the frame; when sampling is needed, calculating a visual angle point position corresponding to an observation point according to needs, and performing complete sample collection on a target by traversing each frame of the panoramic video and the visual angle point of the target needing sample collection on each frame, so that a data set of the target is obtained; performing the operation on any target once, so that sample pictures of all scenes possibly required by the target can be obtained, so that the complete picture set of the target sample can be collected in a short time.

Description

technical field [0001] The present invention relates to the field of artificial intelligence research, in particular to the field of image sample data set construction, and in particular to a construction method, device and terminal equipment of an image sample data set based on dynamic panoramic image reconstruction. Background technique [0002] For a long time, image sample data has been the cornerstone of the computer to perform target detection and recognition tasks through images, whether it is the traditional methods of image feature extraction + machine learning algorithms, or the emerging methods of training various convolutional neural networks. The methods of deep learning all rely on the collection of image sample data. In the process of realizing the present invention, the inventors found that there are at least the following problems in the prior art: in the process of algorithm research such as target detection and recognition, it is necessary to prepare corre...

AI Technical Summary

Problems solved by technology

However, for some special scenarios, specific targets must be sampled in detail. If traditional sample collection methods are used, samples need to be collected many times. The workload is huge, and the obtained sample data sets may not be complete. ; If some scenes are missed, re-sampling is required; sometimes limited by the actual situation on site, the cost of multiple sampling is high, and it may not be able to complete all the missing scenes

Images