| 摘要 |
A learning method for object detection comprises: (a-1) a step of converting a feature point of an image into a binary value (j); (b-1) a step of using the binary value (j) of step (a-1) to calculate a sorting ability value (a_j) which is an ability for sorting when the corresponding binary value (j) occurs; (c-1) a step of sorting the sorting ability (a_j) calculated in step (b-1); and (d-1) a step of calculating a cumulative sum (c_i) of the sorted value (s_i) of the sorting ability (a_j) sorted in step (c-1). According to the learning method for object detection and an object detection method, a memory capacity required for compressing and restoring learning data can be reduced. |
