A method, system and electronic device for inserting image frames between videos

Abstract

The present application relates to a method, system and electronic equipment for inserting image frames between videos. Including: before the video loss time and after the video recovery time, respectively select m frames containing pedestrian feature maps, and collect a set number of human body posture points from each feature map; input all human body posture points into the Alex Net network, The Alex Net network uses the method of combining cubic polynomial fitting and cubic spline interpolation to predict the pedestrian posture of the image to be restored; the human body posture point corresponding to the feature map before the video loss time is input into the LSTM network, and the image to be restored is obtained. Pedestrian attitude prediction result; obtain the image to be restored according to the pedestrian attitude prediction result of the Alex Net network and the LSTM network, and calculate the insertion position of the image to be restored in the video, and insert the image to be restored into the corresponding position in the video position. This application improves the accuracy of existing algorithms.

Description

technical field [0001] The present application belongs to the technical field of frame insertion between videos, and in particular relates to a method, system and electronic equipment for inserting image frames between videos. Background technique [0002] With the application of more and more entertainment-level short videos, there is a large demand for short video transmission and playback in China, but at the same time, the problem is that video transmission will be caused by various reasons such as network transmission. Frames are lost during the process. In this case, the viewer will feel discontinuous in vision, which not only affects the viewing experience, but may also cause the video to have low usability due to the loss of key information. [0003] In the case of frame loss, existing technologies often improve the quality of the transmission process so that the video is uploaded or downloaded without frame loss as much as possible, but there is nothing that can be ...

AI Technical Summary

Problems solved by technology

[0002] With the application of more and more entertainment-level short videos, there is a large demand for short video transmission and playback in China, but at the same time, the problem is that video transmission will be caused by various reasons such as network transmission. Frames are lost during the process. In this case, the visual discontinuity of the viewer will be caused, which not only affects the viewing experience, but may also cause the video to have low usability due to the loss of key information.

The literature [Walker J, Marino K, GuptaA, et al.The pose knows: Video forecasting by generating pose futures[C] / / Proceedings of the IEEE International Conference on Computer Vision.2017:3332-3341.] shows that under the prediction In terms of the possible behavior of a frame, the VAE method [Kingma D P, WellingM.Auto-encoding variational bays[J].arXiv preprint arXiv:1312.6114,2013.] can be used for human pose prediction, but these pose predictions are only for future frames Some predictions of , do not involve using these predictions to complete two related videos but with missing frames in the middle

Images