yyfyan
yyfyan
I had trained the semantic segmentation model on BDD100K,now I want to test the model on the test set. How to test it? @fyu
ok,thank you. I have solved the issues.
And, Can you add the improved ResNet version? original ResNet: 7*7 convolution improved ResNet: 7*7 convolution ---> 3*3 convolution+3*3 convolution+3*3 convolution
sorry,I don't find the url too.
the improved ResNet version is used in PSPNet. https://github.com/hszhao/PSPNet
I used the evalpyt2.py [evalpyt2.py.zip](https://github.com/isht7/pytorch-deeplab-resnet/files/2212917/evalpyt2.py.zip)
and when read the train.py，I found the flip augment is not used,becauses you only flip the image ,not use the double images(origali image and flip image).
you meet the follow problem? 
@Leotju
Hi, best work, I have a question about whether checkerboard images can be used for multiple camera calibration. Beacuse of we need calibrate image under the water, and the random...