the confusion of trainning result

[jaminChen999]

I prepared my datasets and resized image to 768 x 512 for training to car and pedestrian network's model simutaneously,then the result's is that pedestrian's detection is good,but the majority bigger car's detection is not good. I am confused for this result,there some trainning parameters to be changed,Can you help me ,thanks!

[ChenLi2013]

hi jaminChen999 , how about the time of rrc at the resolution of 768x512

[shengtao96]

Hi jaminChen999, how do you modify the model to fit the different resolution?
Change Line 69 resize_width = 768
Line 70 resize_height = 512 in the examples/car/rrc_kitti_car.py?

[jaminChen999]

yes， | | 陈志明 邮箱：[email protected] | 签名由 网易邮箱大师 定制 On 01/15/2018 11:32, shengtao96 wrote: Hi jaminChen999, how do you modify the model to fit the different resolution? Change Line 69 resize_width = 768 Line 70 resize_height = 512 in the examples/car/rrc_kitti_car.py? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.