Tensorboard monitor for recent tensorflow and pytorch about tenet

[jungwon-choi]

plus you have to edit each train.py as below
from

info = {
            'loss': loss,
            'sr_loss': sr_loss,
            'dm_loss': dm_loss,
            'psnr': cur_psnr
       }

to

 info = {
            'loss': loss.item(),
            'sr_loss': sr_loss.item(),
            'dm_loss': dm_loss.item(),
            'psnr': cur_psnr
        }

[jungwon-choi]

However, I found that tensorflow requires a lot of memory just to log the losses.
So, I changed the code with tensorboard for pytorch.

# import tensorflow as tf
from torch.utils.tensorboard import SummaryWriter
import numpy as np

try:
    from StringIO import StringIO  # Python 2.7
except ImportError:
    from io import BytesIO         # Python 3.x


class Tf_Logger(object):

    def __init__(self, log_dir):
        """Create a summary writer logging to log_dir."""
        self.writer = SummaryWriter(log_dir)

    def scalar_summary(self, tag, value, step):
        """Log a scalar variable."""
        self.writer.add_scalar(tag, value, step)

    def image_summary(self, tag, images, step):
        """Log a list of images."""
        for i, img in enumerate(images):
            if len(img.shape) == 2:
                img = img[np.newaxis,:,:,np.newaxis]
            elif len(img.shape) == 3:
                img = img[np.newaxis,:,:,:]
            self.writer.add_image('%s/%d' % (tag, i), img, step)
    
    def histo_summary(self, tag, values, step, bins=1000):
        """Log a histogram of the tensor of values."""

        # Create a histogram using numpy
        counts, bin_edges = np.histogram(values, bins=bins)

        # Fill the fields of the histogram proto
        hist = tf.HistogramProto()
        hist.min = float(np.min(values))
        hist.max = float(np.max(values))
        hist.num = int(np.prod(values.shape))
        hist.sum = float(np.sum(values))
        hist.sum_squares = float(np.sum(values**2))

        # Drop the start of the first bin
        bin_edges = bin_edges[1:]

        # Add bin edges and counts
        for edge in bin_edges:
            hist.bucket_limit.append(edge)
        for c in counts:
            hist.bucket.append(c)

        # Create and write Summary
        self.writer.add_histogram(tag, hist, step)

[jungwon-choi]

In this case, you don't need to transpose RGB images at valid() function.

img = img[0:1, 0:4]
img = torch.clamp(img.cpu()[0, 0], 0, 1).detach().numpy()
gt = torch.clamp(gt.cpu()[0], 0, 1).detach().numpy()
# gt = np.transpose(gt, [1, 2, 0]) 	# permute
output = torch.clamp(output.cpu()[0], 0, 1).detach().numpy()
# output = np.transpose(output, [1, 2, 0])

[guochengqian]

@jungwon-choi thank you for your updates! Really helpful.