Plot with more of 3 channels
luissalgueiro opened this issue · 4 comments
Hello @cheind
I installed your package and tried to customize the scatter example, but I could not plot more than 3 subplots. For example, adding more than 3 colors it only the first 3.
How can I solve the issue?
Thanks
Luis
hard to see what's wrong, share some minimal code please
Hello @cheind, thanks for the fast replay.
The idea is to see every channel of a Tensor with shape [Batch, Height, Widht, Channels] where the number of channels >=4
// number of channels for the tensor to be plotted
channels = 4
with tf.Session(graph=tf.Graph()) as sess:
@tfmpl.figure_tensor
def draw_scatterss(scaled, channels):
'''Draw scatter plots. One for each color.'''
figs= tfmpl.create_figures(channels, figsize=(7,7))
for idx, f in enumerate(figs):
ax = f.add_subplot(111)
ax.imshow(scaled[:,:,idx], cmap='gray')
return figs
aa = tf.placeholder(tf.float32,[14,14,None]) # or')
image_tensor = draw_scatterss(aa, channels)
image_summary = tf.summary.image('scatter', image_tensor)
all_summaries = tf.summary.merge_all()
os.makedirs('log', exist_ok=True)
now = datetime.now()
logdir = "log/" + now.strftime("%Y%m%d-%H%M%S") + "/"
writer = tf.summary.FileWriter(logdir, sess.graph)
#create a tensor with random noise
aa1 = np.random.random((14,14,channels))
summary = sess.run(all_summaries, feed_dict={aa: aa1 })
writer.add_summary(summary, global_step=0)
writer.flush()
Hi @cheind,
I think I have the solution. Only add channels to the max_outputs in the summary
image_summary = tf.summary.image('scatter', image_tensor, max_outputs=channels)
Great code by the way, I learned a lot.
Thanks
Great, closing!