Unmanned aerial vehicle video aesthetic quality evaluation method based on multi-modal deep learning

Abstract

The invention provides an unmanned aerial vehicle video aesthetic quality evaluation method based on multi-modal deep learning, and the method comprises the steps: building an unmanned aerial vehiclevideo aesthetic evaluation data set, carrying out the analysis of an unmanned aerial vehicle video through a multi-modal neural network, extracting high-dimensional features, and carrying out the fusion, thereby achieving the evaluation of the aesthetic quality of the unmanned aerial vehicle video. The method mainly comprises four steps: 1, establishing an unmanned aerial vehicle video aesthetic evaluation data set, and dividing the data set into a positive sample and a negative sample according to video shooting quality; 2, recovering the flight path of the unmanned aerial vehicle and reconstructing a sparse scene three-dimensional structure by using an SLAM technology; 3, respectively extracting features of the input unmanned aerial vehicle video on an image branch, a motion branch and astructure branch through a multi-modal neural network; and 4, fusing the features on the plurality of branches to obtain a final video aesthetics label and a final video scene type. Experiments provethat the method has feasibility, accuracy and universality and can be used for aesthetic evaluation, shooting track recommendation and the like of unmanned aerial vehicle videos.

Description

technical field [0001] The invention relates to a multi-modal deep learning-based aesthetic quality evaluation method for UAV video, which establishes a UAV video aesthetic evaluation data set, analyzes UAV video and extracts high-dimensional features through a multi-modal neural network, Then fusion, so as to realize the evaluation of the aesthetic quality of UAV video, has certain effectiveness and versatility, and belongs to the field of computer vision. Background technique [0002] With the rapid popularization of cameras and smartphones, perception and understanding of visual content has become a research direction in the fields of computer vision and computer photography. Image and video aesthetic quality evaluation is a branch of visual content perception and understanding. Image and video aesthetic quality evaluation aims to use computers to simulate human perception and cognition of beauty, and automatically evaluate the beauty of images and videos. In recent yea...

AI Technical Summary

Problems solved by technology

This method only evaluates the aesthetics of images, and cannot evaluate the aesthetic quality of videos captured by drones.

Images