shekkizh FCN.tensorflow issues

shekkizh / FCN.tensorflow

Tensorflow implementation of Fully Convolutional Networks for Semantic Segmentation (http://fcn.berkeleyvision.org)

MIT License

1.25k stars 527 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

wimdows

#80 CCTV-lulu closed 6 years ago
1
I get nan during the inference stage using my own data ?

#79 Deephome closed 6 years ago
1
MemoryError

#78 MandyMo closed 6 years ago
1
When visualized,I got outputs with black photos ,no results. Why?

#77 SpadesQ closed 6 years ago
7
How to test ?

#76 XZNWU closed 6 years ago
6
% (np_val.shape, subfeed_t.name, str(subfeed_t.get_shape()))) ValueError: Cannot feed value of shape (224, 224) for Tensor 'conv2d_2_input:0', which has shape '(?, 224, 224, 3)'

#75 muradalqurishee closed 6 years ago
10
Question: Could I ask why in the FCN structure.. it looks like using VGG to compute the feature first then input the feature to conv 3 times then deconv 3 times.. I think it might be look like VGG layers then deconv such number of layers

#74 wuqinhao closed 6 years ago
1
why you read annotation only one channel,Not three channels??

#73 liuzhisheng1226 closed 6 years ago
2
Question: why use "np.transpose()" to change the weights when you initialize the weights of conv layers? when I read others code(https://github.com/MarvinTeichmann/tensorflow-fcn/blob/master/fcn8_vgg.py), I find they don't transpose the vgg16-weights. Note that the his vgg16-weights is equal with your used weights except the weights of conv1_1. So I feel confused whether it is necessary to transpose the vgg16-weights when initializing the weights of conv?

#72 OREAw closed 6 years ago
1
image = tf.placeholder(tf.float32, shape=[None, IMAGE_SIZE, IMAGE_SIZE, 6]).

#71 MLjian closed 6 years ago
2
Question: How to input my own picture and get the result?

#70 WANGYuSCU closed 6 years ago
4
Why set the NUM_OF_CLASSESS = 151

#69 MZ59 closed 6 years ago
2
different between annotation and valid_images

#68 engahmed1190 closed 6 years ago
1
FCN32, FCN16?

#67 lily10086 closed 6 years ago
3
why must download imagenet-vgg-verydeep-19.mat?

#66 lily10086 closed 6 years ago
3
How to load model one time but inference more time ?

#65 daicoolb closed 6 years ago
3
Training accuracy 100%, validation accuracy half of it (near 50%); blank prediction image

#64 sanjeevan1234 closed 6 years ago
4
unsupported pickle protocol 4

#63 ShiningRuby closed 6 years ago
1
run error

#62 haddis3 closed 6 years ago
4
Pretrained model

#61 Zumbalamambo closed 6 years ago
1
need trained weights

#60 rnunziata closed 6 years ago
1
Can you show me, pixel accuracy you retrieve when you evaluate your model in entire validate set ? thanks you so much!

#59 ghost closed 6 years ago
2
put training and validation loss in the same graph in tensorboard

#58 txzhao closed 6 years ago
3
up sampling phase structure?

#57 bhralzz closed 6 years ago
1
A question in Image_Cmaped.ipynb

#56 fanhualta closed 6 years ago
2
Windows GPU 0.12version issue

#55 fanhualta closed 7 years ago
1
Getting blank predicted images

#54 NitishMutha closed 6 years ago
13
How to test and demo?

#53 zhangzhizz closed 6 years ago
9
Why are you padding on the sixth convolutional layer?!

#52 TheRevanchist opened 7 years ago
2
ValueError: Cannot feed value of shape (0,) for Tensor 'input_image:0', which has shape '(?, 224, 224, 3)'

#51 jabgle closed 7 years ago
9
Why Accuracy and Loss are not corresponding?

#50 dlrPHS2 closed 6 years ago
2
Update TensorflowUtils.py

#49 fndxy closed 7 years ago
0
Why the scale factor in caffemodel is ignored in tf equivalent? Should we scale the input to 1/256 or 1/128 as caffe used to do?

#48 Edwardmark closed 7 years ago
2
The order of the fc layers

#47 Edwardmark closed 6 years ago
2
relu5_3

#46 yossitsadok closed 7 years ago
2
GPU training Nan loss

#45 polybahn closed 6 years ago
2
higher order of channel?

#44 bhralzz closed 7 years ago
1
Test the trained model?

#43 bhralzz closed 7 years ago
4
training termination

#42 bhralzz closed 7 years ago
1
how to decrease space used for logging?

#41 bhralzz closed 7 years ago
4
How to visualize the prediction?

#40 Dean-TianZhang closed 7 years ago
5
How can I train arbitrary sized images?

#39 tianjihx closed 7 years ago
2
Slight modification, large loss values?

#38 grey-area closed 6 years ago
2
Inference using large amount of GPU memory

#37 NeuroRoboticTech closed 7 years ago
1
some problem about the result(output map)

#36 yilinchenXD closed 7 years ago
1
using --mode=visualize after trained the model, but I got the wrong pred_x.png

#35 ChinaYi closed 7 years ago
11
Test with single image

#34 Yaredoh closed 7 years ago
2
How to calculate the segmentation accuracy of the model？

#33 yeshenlin closed 7 years ago
2
how can I use colored images annotation

#32 Yaredoh closed 7 years ago
4
Clean the checkpoint after any modify

#31 RichardWang12 closed 7 years ago
0

Previous Next