gwding/draw_convnet

draw_convnet's Issues

Modifying for 1-D convolutions

Hi @gwding

Firstly, thanks for the code! I am trying to modify it for 1-D layers. I am able to get the layers in thin rectangles in order to represent, but I am not able to shift the mappings for these new layers. Could you please guide me about how to do it?

Attached is what I am able to get.

Thanks.

How to make 3 Fully connected layer

Hi, thanks for the code. However for my problem I have 3 fully connected layer, however when change
size_list = [(fc_unit_size, fc_unit_size)] * 4 , num_list = [4500, 4500,4500, 1148] I have this error _IndexError: index 3 is out of bounds for axis 0 with size 3

Do you know how to overcome this ?
Thanks !

Not able to add layers

I need to add more convolution layers in my model. But I am not able to understand the code to do so. Can you please help me to do so?

Below is the visualization of model which I want to make. Please help.

Layer (type) Output Shape Param #

conv2d_11 (Conv2D) (None, 126, 126, 16) 448

---

conv2d_12 (Conv2D) (None, 124, 124, 32) 4640

---

max_pooling2d_9 (MaxPooling2 (None, 62, 62, 32) 0

---

conv2d_13 (Conv2D) (None, 60, 60, 32) 9248

---

max_pooling2d_10 (MaxPooling (None, 30, 30, 32) 0

---

conv2d_14 (Conv2D) (None, 28, 28, 32) 9248

---

max_pooling2d_11 (MaxPooling (None, 14, 14, 32) 0

---

conv2d_15 (Conv2D) (None, 12, 12, 32) 9248

---

max_pooling2d_12 (MaxPooling (None, 6, 6, 32) 0

---

dropout_5 (Dropout) (None, 6, 6, 32) 0

---

flatten_3 (Flatten) (None, 1152) 0

---

dense_5 (Dense) (None, 32) 36896

---

dropout_6 (Dropout) (None, 32) 0

---

dense_6 (Dense) (None, 3) 99

Total params: 69,827
Trainable params: 69,827
Non-trainable params: 0

Other architectures?

Hi and thanks a lot for your effort !

It would be nice if other contributors add other examples (other than convnet) based on the code.

I was looking for someone who has a code for drawing a sketch of convolutional autoencoders, in the end, I decided to do it starting from draw_convnet.py

Regards.

Layer overlap

When the size of the layer is too big, the following layer will cover part of the previous layer, like the first two feature maps in this figure.

I suggest that the line 167 should be changed to
for ind in range(len(size_list)-1,-1,-1):
It works like:

How to use the code?

Need more examples or ideas on how to use the code to make a new visualization from an existing model in our hand

Hi, I have a problem, maybe you can answer me, thanks

How do I increase picture size

Needs a more comprehensive use of the library in the README

Issues arising from the lack of examples on how to use in the README file

How to remove one fully connected layer?

I only have one fully connected layer in my conv net. However, when I try to remove it from list num_list = [1, 500] I get add_layer(patches, colors, size=size_list[ind], num=num_show_list[ind], IndexError: list index out of range How do I solve this problem?

Vector image

I guess saving figures as vector graphics would have a better resolution:

fig_ext = '.eps'
fig.savefig(os.path.join(fig_dir, 'convnet_fig' + fig_ext),
      bbox_inches='tight', pad_inches=0, format='eps')

Layer Overlap

Please help me to properly arrange layers and corresponding text

TypeError: 'map' object is not subscriptable

runfile('S:/draw_convnet-master/draw_convnet.py', wdir='S:/draw_convnet-master')
Traceback (most recent call last):

File "", line 1, in
runfile('S:/draw_convnet-master/draw_convnet.py', wdir='S:/draw_convnet-master')

File "C:\Anaconda3\lib\site-packages\spyderlib\widgets\externalshell\sitecustomize.py", line 714, in runfile
execfile(filename, namespace)

File "C:\Anaconda3\lib\site-packages\spyderlib\widgets\externalshell\sitecustomize.py", line 89, in execfile
exec(compile(f.read(), filename, 'exec'), namespace)

File "S:/draw_convnet-master/draw_convnet.py", line 128, in
num=num_show_list[ind],

TypeError: 'map' object is not subscriptable

I use python 3.5.2