Semantic_Segmentation/plot_examples.py at main · p-rosit/Semantic_Segmentation · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
import os
import shutil
import argparse
import importlib
import tqdm

import torch
import numpy as np
import matplotlib.pyplot as plt


def parse_command_line():
    """
    Parse command line and get the name of the experiment and type
    """
    parser = argparse.ArgumentParser()
    parser.add_argument('-e', '--exp', type=str, help='experiments name')
    parser.add_argument('-imgs', '--n', type=int, help='Amount of pictures to make')
    parser.add_argument('-best','--best', type=bool, default=False, help='Use best checkpoint')
    parser.add_argument('-rm','--remove',type=bool,default=False, help='Remove previously generated images in folder')
    args = parser.parse_args()
    return args


def load(config):
    opt = config['opt']

    resume = os.path.join('exp', opt.exp)
    if opt.best:
      resume_file = os.path.join(resume, 'best_checkpoint.pt')
      print("=> loading best checkpoint '{}'".format(resume))
    else:
      resume_file = os.path.join(resume, 'checkpoint.pt')
      print("=> loading checkpoint '{}'".format(resume))

    if os.path.isfile(resume_file):
        checkpoint = torch.load(resume_file, map_location=config['device'])

        # print(checkpoint['inference'])
        # error(':)')

        config['inference']['net'].load_state_dict(checkpoint['state_dict'])
        config['train']['optimizer'].load_state_dict(checkpoint['optimizer'])
        config['train']['epoch'] = checkpoint['epoch']
        config['inference']['best_loss'] = checkpoint['best_loss']
        print("=> loaded checkpoint '{}' (epoch {})".format(resume, checkpoint['epoch']))
    else:
        print("=> no checkpoint found at '{}'".format(resume))
        exit(0)


def save_picture(config, output, label):
    _, pred = torch.max(output[-1], dim=0)

    plt.subplot(2,1,1)
    plt.imshow(pred.cpu())
    plt.xticks([])
    plt.yticks([])
    plt.title('Network Performance')

    plt.subplot(2,1,2)
    plt.imshow(label.squeeze().cpu())
    plt.xticks([])
    plt.yticks([])
    plt.title('Ground Truth')

    plt.savefig(os.path.join(config['opt'].pic_path, config['phase'] + "%02d"%(config['n']) + ".png"), bbox_inches='tight')
    plt.close()


def make_picture_minibatch(config, phase, inputs, labels):
    outputs = config['inference']['net'].model(inputs)

    for i in range(outputs.shape[0]):
      if config['n'] > config['opt'].n:
        break

      save_picture(config, outputs[i], labels[i])

      config['n'] += 1

    return config


def make_pictures(config, data_func, phase):
    dataloader = data_func(phase)

    config['n'] = 0
    config['phase'] = phase
    for _, (inputs, labels) in tqdm.tqdm(enumerate(dataloader), total=np.ceil(config['opt'].n/len(dataloader)).astype('int'), ascii=True):

      make_picture_minibatch(config, phase, inputs, labels)

      if config['n'] >= config['opt'].n:
        break


def init():
    opt = parse_command_line()
    exp_path = os.path.join('exp', opt.exp)
    pic_path = os.path.join(exp_path, 'pictures')
    opt.exp_path = exp_path
    opt.pic_path = pic_path

    try: os.makedirs(pic_path)
    except FileExistsError: pass

    for filename in os.listdir(pic_path):
      file_path = os.path.join(pic_path, filename)
      try:
          if os.path.isfile(file_path) or os.path.islink(file_path):
              os.unlink(file_path)
          elif os.path.isdir(file_path):
              shutil.rmtree(file_path)
      except Exception as e:
          print('Failed to delete %s. Reason: %s' % (file_path, e))

    config = importlib.import_module('hyperparameters').__config__
    network = importlib.import_module('models.network')
    config['opt'] = opt
    config['data_provider'] = importlib.import_module(config['data_provider'])
    config['device'] = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
    print("=> Using", config['device'])

    func = network.make_network(config)
    load(config)

    return config


if __name__ == "__main__":
    config = init()
    print("=> Loading data")
    data_func = config['data_provider'].init(config)

    for phase in ['train','valid','test']:
      print("Start " + phase + " " + config['opt'].exp)
      make_pictures(config, data_func, phase)

    print("=> Completed")