-
Hello I am trying to train hr_res101('train') with Wider face data. I compiled a CPU only version with matlab 2016b and matconvnet in ubuntu 14.04. After running the hr_res101('train') script I am get…
-
my output symbol file like this, and this file can not load by mx.model.load_checkpoint()
@tornadomeet
@qiyuangong
@bittnt
{
"nodes": [
{
"op": "null",
"name": "data",
…
-
-
First for the installation it says to install everything in "requirements.txt" but file is missing.
Also, I am trying to test your code but I need the "prostage-cityscapes-fcn8s.prototxt". In gener…
-
Hi,
May I ask that when you update FCN8s or FCN 16s?
Thanks!
-
Hey ,
I am trying to use opencv_contrib with dnn module to use PASCAL VOC caffemodel.
I have a set up of,
A working Opencv , opencv_contrib along with dnn module .
I have used caffe_googlenet.cpp …
-
Hi Peiyun,
Thank you so much for sharing with us your awesome work!
While I've been experimenting with training this net and analyzing your train code, a few points have come up on which i would be …
-
I found that when you trained the FCN8s model, the learning rate is too small (1e-14). I remember that the learning rate is set to 1e-4 in the original FCN paper. I am a little confused.
Can you gi…
-
model: fcn8s
default parameter, n_epoch=50
when epoch=2,mean IoU=0.19
epoch=5,meanIoU=0.425
epoch=6,7,8....49,mean IoU acc and other metrics unchanged
Has someone trained …
jetxa updated
6 years ago
-
All three classes fcn8s fcn16s fcn32s has padding of 100 on the first conv layer, why use such a large padding?