It uses the object template(festures) as well as the current image.
It uses the rpn of faster rcnn.
'Papers > Machine learning' 카테고리의 다른 글
| CornerNet: Detecting Objects as Paired Keypoints (0) | 2019.06.22 |
|---|---|
| Generative adversary network (0) | 2019.06.21 |
| Tracking without bells and whistles (0) | 2019.06.07 |
| zero skip (0) | 2019.06.04 |
| Backpropagation (0) | 2019.05.30 |