Distilling-Object-Detectors

Distilling-Object-Detectors copied to clipboard

为什么Teacher模型是1G大小，Student也是1G大小，这个蒸馏的意义在哪里

DonMuv

您好，请问您是否遇到了这样的问题，当使用mGPUs时，明明显示可使用的卡有多张，但最终模型只加载到了一张卡上，希望您可以解答一下。

zengbohan0217

Hi, where can I get res101h pretrained model

TheLostIn

about imitation loss weight

6

Thanks for your code ! When i use imitation loss in my dataset and work, i'm confused about how to determine the imitation loss weight, without imitation loss, my total...

chongkuiqi

xiweiya

can't load model!

5

 您好，我按照给定命令训练的时候，为什么到这一步就卡住了？没有任何反应和提示。 当我不加参数运行的时候，说缺少ｖgg16_caffe.pth，然后我下载了一个，放在里面，又出现了下面的情况。  请问teacher不是vgg16吗？为甚么使用ｖgg11_caffe.pth　pretrained model

oxygenman

problem with installing

2

Hi, I. tried to install the program, then there was a problem with nms_cuda.c, the file seems to be missing, is there anything particular that I have to do before...

pipilurj

ImportError: torch.utils.ffi is deprecated. Please use cpp extensions instead.

1

It seems that the build part for models is a bit old. With new Pytorch the ffi API is deprecated and no way to adapt it to cpp extension unless...

mmkamani7

when distillation with FPN，which feature map shuold we select to calculate imitation loss, feature map of FPN(P3、P4...) or feature of backbone(C3、C4...) or others?

4

UcanSee

Is res50h pertained on ImageNet?

2

In your experiment using res50h model, do you use image-net pretrained weights or randomly initialized weights?

shuzhangcasia