Hi -- what is meant by this line in the paper?
In addition, as the minimum scale of the objects in the KITTI dataset is much smaller than the original configuration, we adjusted the corresponding scale of conv4_3 from 0.1 to 0.066. We also removed the last global pooling layer of the original SSD model and set the scale of conv 10_2 to 0.85.
Hi -- what is meant by this line in the paper?
In addition, as the minimum scale of the objects in the KITTI dataset is much smaller than the original configuration, we adjusted the corresponding scale of conv4_3 from 0.1 to 0.066. We also removed the last global pooling layer of the original SSD model and set the scale of conv 10_2 to 0.85.Is it reflected in the code?