tools.py

import os
import glob
import argparse
from tqdm import tqdm
import torch
import torch.nn as nn
from torchvision import datasets, transforms
from torch.autograd import Variable
import numpy as np
from skimage import io
import cv2

USE_CUDA = torch.cuda.is_available()
DEVICE = 0
def to_cuda(v):
    if USE_CUDA:
        return v.cuda(DEVICE)
    return v

def accuracy(input, target):
    return 100 * float(np.count_nonzero(input == target)) / target.size

def conf_m(output, target_th):

  output_conf=((output.data.squeeze()).transpose(1,3)).transpose(1,2)
  output_conf=(output_conf.contiguous()).view(output_conf.size(0)*output_conf.size(1)*output_conf.size(2), output_conf.size(3))
  target_conf=target_th.data.squeeze()
  target_conf=(target_conf.contiguous()).view(target_conf.size(0)*target_conf.size(1)*target_conf.size(2))
  return output_conf, target_conf

def write_results(ff, save_folder, epoch, train_acc, test_acc, change_acc, non_ch, train_losses, val_losses):
    ff=open('./' + save_folder + '/progress_run.txt','a')
    ff.write('train: ')
    ff.write(str('%.3f' % train_acc))
    ff.write(' ')
    ff.write(' val: ')
    ff.write(str('%.3f' % test_acc))
    ff.write(' ')
    ff.write(' CHANGE: ')
    ff.write(str('%.3f' % change_acc))
    ff.write(' ')
    ff.write(' NON_CHANGE: ')
    ff.write(str('%.3f' % non_ch))
    ff.write(' ')
    ff.write(' E: ')
    ff.write(str(epoch))
    ff.write('         ')
    ff.write(' TRAIN_LOSS: ')
    ff.write(str('%.3f' % train_losses))
    ff.write(' VAL_LOSS: ')
    ff.write(str('%.3f' % val_losses))
    ff.write('\n')