DFN-tensorflow

DFN-tensorflow copied to clipboard

the result of training

18

Why focal_loss and mean_iou are always zero when the training progress is going on? I didn't change the model, I used pascalvoc2012 for training

gaowu9595

yangrui51

Bumps [tensorflow-gpu](https://github.com/tensorflow/tensorflow) from 1.2.0 to 1.15.2. Release notes *Sourced from [tensorflow-gpu's releases](https://github.com/tensorflow/tensorflow/releases).* > ## TensorFlow 1.15.2 > # Release 1.15.2 > > ## Bug Fixes and Other Changes > *...

dependabot[bot]

I have a issue about how to use the supervisory signal. The paper said "The supervisory signal of the network is obtained from the semantic segmentation’s groundtruth with a traditional...

Z-Ianthe

binary segmentation dataset

1

Hi`～,thanks for sharing the code for this project ,but u said that only the problem of binary segmentation can be solved,where can I find binary segmentation dataset, I searched for...

zhenmafan7

about "random.sample()"

1

In file utils , I found the function get_batch_of__trainval() uses random.sample to get batch_list in line 65, but random.sample is a pseudorandom function. I think it may influence the final...

myknowntime

mathpopo

Hi. Thanks for the awesome work. The paper is really great and it makes so much sense. Is it possible to get the trained weight files for the binary segmentation...

calicratis19

Missing parameter(category = "test") for the get_batch_of_test function in utils.py file

3

1. I was able to train a model on the kitti dataset from your code but while testing i got a name error: Category not defined. When i checked with...

narendoraiswamy

About pretrained ResNet101?

1

Hi, thanks for your great work. Why don't you load ImageNet pretrained ResNet101 model parameters in the code?

FightForCS