METHOD OF CARTOON STYLE SUMMARY FOR VEDIO USING DEEP LEARNING

摘要

A first step of dividing an image for each camera scene and extracting a plurality of frames for each camera scene; A second step of calculating scores for the frames by inputting the frames into a dedicated convolutional neural network (CNN) model; A third step of selecting the remaining frames so that a preset number of frames is satisfied based on the frame having the highest score; A fourth step of converting the selected frame into a cartoon style; And a fifth step of arranging the converted frames according to the cartoon layout in chronological order.