LEARNING METHOD AND LEARNING DEVICE FOR DETECTING OBJECT EXISTENCE BY DETECTING BOTTOM LINES AND TOP LINES OF NEAREST OBSTACLES, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

摘要

To provide a learning method and a learning device for detecting object existence by detecting bottom lines and top lines of nearest obstacles.SOLUTION: A method is characterized in that a learning device includes: (a) a step of causing a first CNN to generate first encoded feature maps and first decoded feature maps, and causing a second CNN to generate second encoded feature maps and second decoded feature maps; (b) a step of generating, respectively, first and second obstacle segmentation results in which where bottom lines and top lines line are located for each column is estimated, by referring to the first and the second decoded feature maps, respectively; (c) a step of estimating object existence by referring to the first and the second obstacle segmentation results; (d) a step of generating losses by referring to the object existence and their corresponding ground truth images; and (f) a step of backpropagating the losses, to thereby learn parameters of the CNNs.SELECTED DRAWING: Figure 11