LEARNING METHOD AND LEARNING DEVICE FOR REMOVING JITTERING ON VIDEO ACQUIRED THROUGH SHAKING CAMERA BY USING A PLURALITY OF NEURAL NETWORKS FOR FAULT TOLERANCE AND FLUCTUATION ROBUSTNESS IN EXTREME SITUATIONS AND TESTING METHOD AND TESTING DEVICE USING THE SAME

摘要

Provides for Fault Tolerance and Fluctuation Robustness under extreme conditions, using a Neural Network to remove jitter on video using a video generated by a shaking camera. A method of detecting turing, comprising: generating, by a computing device, each t-th mask corresponding to each object in a t-th image; For each t-th mask, each t-th cropped image, each t-1 th mask and each t-1 th cropped image, the second neural network operation is applied at least once to generating a respective t-th object motion vector of each object pixel included in the image; and generating each t-th jittering vector corresponding to each reference pixel among pixels in the t-th image with reference to each t-th object motion vector, whereby the present invention comprises: It can be used for video stabilization, ultra-precise object tracking, behavior prediction, and motion decomposition.