Tensorflow-SegNet

Tensorflow-SegNet copied to clipboard

Hi.. Thank you for the code.. I can successfully run this code on CamVid dataset and now I want to try it on my dataset. My images are of **integer...

amitadhiman

Output_Image

2

Is there currently a way to see all of the segmented test images? By default only one of these images is saved, which is the testing_image.png. And in model.py, I...

tom-bu

Upsampling with pool indices

2

Hi, Thanks for posting such useful code! I noticed that that the upsampling doesn't seem to use the indices from the max-pool as they do in the SegNet paper. Have...

gspell

Nan in summary histogram for: conv1/biases_1

1

Caused by op u'conv1/biases_1', defined at: File "main.py", line 52, in tf.app.run() File "/home/gis322/anaconda2/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run _sys.exit(main(_sys.argv[:1] + flags_passthrough)) File "main.py", line 49, in main model.training(FLAGS, is_finetune=False) File...

flyking1994

@tkuanlun350 Apparently, `Road` and `Road_marking` have been merged into one (https://github.com/alexgkendall/SegNet-Tutorial/issues/51). Hence, after the merge of PR #22, the `Road` shows the colour of `Road_marking` and every label shows the...

americast

Hello, I have image having dimension 1500*1500. I want to randomly crop to 256*256 and feed it to this model for training. Also, i have only two class. What should...

sanjeevan1234

No function definition for upsample_with_pool_indices

1

In model.py file, no definition of upsample_with_pool_indices() is given. Can anyone provide it?

Anabik

ddoe777

Error in training images with larger size like 1440x960

1

Try to train with my own data, follow the step here #5 GPU ran out of memory, 12 GB GPU. Any idea for the limitation of the input image size?...

troylhy1991

Skip connections?

2

Hello, it's great work showing how to use conv_transpose. Do you think the skip connections between encoder layer and decoder layer help? Some papers use it others not (your case)....

joyousrabbit