Scribd
Upload
9 views2 pages

Python Image Processing Utilities

The document contains Python code for various utility functions and classes related to neural network initialization, image loading, and logging. Key functions include custom weight initialization for neural network layers, directory creation, and a Logger class for logging metrics. The code also includes assertions for equality checks and a method to load image files from a specified folder.

Uploaded by

kangja1530
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
9 views2 pages

Python Image Processing Utilities

The document contains Python code for various utility functions and classes related to neural network initialization, image loading, and logging. Key functions include custom weight initialization for neural network layers, directory creation, and a Logger class for logging metrics. The code also includes assertions for equality checks and a method to load image files from a specified folder.

Uploaded by

kangja1530
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd

from __future__ import print_function

import errno
import os
import numpy as np
# from PIL import Image
import torch
import [Link] as nn

EPS = 1e-7

def assert_eq(real, expected):


assert real == expected, '%s (true) vs %s (expected)' % (real, expected)

def assert_array_eq(real, expected):


assert ([Link](real-expected) < EPS).all(), \
'%s (true) vs %s (expected)' % (real, expected)

def load_folder(folder, suffix):


imgs = []
for f in sorted([Link](folder)):
if [Link](suffix):
[Link]([Link](folder, f))
return imgs

# def load_imageid(folder):
# images = load_folder(folder, 'jpg')
# img_ids = set()
# for img in images:
# img_id = int([Link]('/')[-1].split('.')[0].split('_')[-1])
# img_ids.add(img_id)
# return img_ids

# def pil_loader(path):
# with open(path, 'rb') as f:
# with [Link](f) as img:
# return [Link]('RGB')

def weights_init(m):
"""custom weights initialization."""
cname = m.__class__
if cname == [Link] or cname == nn.Conv2d or cname == nn.ConvTranspose2d:
[Link].normal_(0.0, 0.02)
elif cname == nn.BatchNorm2d:
[Link].normal_(1.0, 0.02)
[Link].fill_(0)
else:
print('%s is not initialized.' % cname)

def init_net(net, net_file):


if net_file:
net.load_state_dict([Link](net_file))
else:
[Link](weights_init)

def create_dir(path):
if not [Link](path):
try:
[Link](path)
except OSError as exc:
if [Link] != [Link]:
raise

class Logger(object):
def __init__(self, output_name):
dirname = [Link](output_name)
if not [Link](dirname):
[Link](dirname)

self.log_file = open(output_name, 'w')


[Link] = {}

def append(self, key, val):


vals = [Link](key, [])
[Link](val)

def log(self, extra_msg=''):


msgs = [extra_msg]
for key, vals in [Link]():
[Link]('%s %.6f' % (key, [Link](vals)))
msg = '\n'.join(msgs)
self.log_file.write(msg + '\n')
self.log_file.flush()
[Link] = {}
return msg

def write(self, msg):


self.log_file.write(msg + '\n')
self.log_file.flush()
print(msg)

You might also like