search

Cli98 / DMNet

Official implementation for DMNet: Density map guided object detection in aerial image (CVPR 2020 EarthVision workshop)
https://openaccess.thecvf.com/content_CVPRW_2020/papers/w11/Li_Density_Map_Guided_Object_Detection_in_Aerial_Images_CVPRW_2020_paper.pdf
93 stars 14 forks source link

about train probelm? #2

Closed weqvr closed 3 years ago

weqvr commented 3 years ago

train the detection network , you use validation set or train set?

Cli98 commented 3 years ago

@weqvr

Hi and I use training set. It's the same as in the dataset of Visdrone 2018.

weqvr commented 3 years ago

so you use the whole images in train set to train the detection network, and test the crop image with the trained detection net ework？ Is right？

Cli98 commented 3 years ago

so you use the whole images in train set to train the detection network, and test the crop image with the trained detection net ework？ Is right？

@weqvr No. I use whole images + density crops as training data. So does the case for validation.

Cli98 commented 3 years ago

Close this issue by now as the thread remains inactive. Feel free to reopen if you need.