LEARNING METHOD AND TESTING METHOD FOR MONITORING BLIND SPOT OF VEHICLE, AND LEARNING DEVICE AND TESTING DEVICE USING THE SAME

摘要

The present invention relates to a learning method of a convolutional neural network (CNN) for monitoring one or more blind spots of a reference car, so that the learning apparatus outputs class information and location information about the vehicle to be observed included in a training image by a detector. to do; By causing the clue information extraction layer to output the clue information about the vehicle to be observed using the output information, and the FC layer for blind spot checking to perform a neural network operation using the clue information or a processed value determining whether the vehicle to be observed is located in the blind spot; and the blind spot loss value obtained by comparing the result of the determination with the first GT corresponding thereto is backpropagated to learn the parameters of the FC layer for checking the blind spot, and the class information and the location information corresponding thereto A learning method comprising a; backpropagating the vehicle detection loss value obtained by comparing with the second GT to learn the parameters of the detector is disclosed.