realtime-style-transfer-Unity/compute_permutation_feature_importance.py at main · Hidden-dreamz/realtime-style-transfer-Unity · 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
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
#!/usr/bin/env python3

"""
This script uses Permutation Feature Importance as introduced here:
Fisher, Aaron, Cynthia Rudin, and Francesca Dominici. “All models are wrong, but many are useful: Learning a variable’s importance by studying an entire class of prediction models simultaneously.” http://arxiv.org/abs/1801.01489 (2018).

Based on an article by Christoph
"""

import sys

import tqdm

from pathlib import Path
import logging
import argparse
import pickle
import datetime
import tensorflow as tf

from realtime_style_transfer.dataloaders import *


class PermutationFeatureImportanceData():
    channel_contributions: dict
    baseline_losses: dict
    num_samples: int

    def __init__(self):
        self.baseline_losses = {loss_name: 0.0 for loss_name in
                                ("loss", "style_loss", "feature_loss", "total_variation_loss", "depth_loss")}
        self.channel_contributions = {loss_name: {n: 0.0 for n, c in config.channels}
                                      for loss_name in self.baseline_losses.keys()}

        self.num_samples = 0
        self.num_samples_and_permutations = 0


log = logging.getLogger()

argparser = argparse.ArgumentParser()
argparser.add_argument('--checkpoint_path', '-C', type=Path, required=True)

args = argparser.parse_args()
checkpoint_path = args.checkpoint_path

from realtime_style_transfer.shape_config import *

num_styles = 1
config = ShapeConfig(hdr=True, num_styles=num_styles)

from realtime_style_transfer.models import styleTransfer, stylePrediction, styleTransferTrainingModel, styleLoss
from realtime_style_transfer.dataloaders import wikiart


def progressify(iterable, **kwargs):
    return tqdm.tqdm(iterable, file=sys.stdout, ascii=False, **kwargs)


cache_dir = Path(__file__).parent / "cache"

permutation_importance_data_cache_file_path = cache_dir / "permutation_feature_importance"

permutation_importance_data = PermutationFeatureImportanceData()
if permutation_importance_data_cache_file_path.exists():
    log.info(f"Loading cached permutation feature importances from {permutation_importance_data_cache_file_path}" + \
             f"(created at {datetime.datetime.fromtimestamp(permutation_importance_data_cache_file_path.lstat().st_ctime)})")
    permutation_importance_data = pickle.load(permutation_importance_data_cache_file_path.open(mode='rb'))
    print(permutation_importance_data.__dict__)

log.info("Loading dataset...")
training_dataset, validation_dataset = wikiart.get_hdr_dataset(config.input_shape,
                                                               batch_size=1,
                                                               output_shape=config.output_shape,
                                                               cache_dir=cache_dir,
                                                               channels=config.channels,
                                                               seed=278992)

style_transfer_training_model = styleTransferTrainingModel.make_style_transfer_training_model(
    style_predictor_factory_func=lambda num_top_parameters: stylePrediction.create_style_prediction_model(
        config.input_shape['style'][1:], stylePrediction.StyleFeatureExtractor.MOBILE_NET, num_top_parameters
    ),
    style_transfer_factory_func=lambda: styleTransfer.create_style_transfer_model(
        input_shape=config.input_shape['content'],
        output_shape=config.output_shape,
        bottleneck_res_y=config.bottleneck_res_y,
        bottleneck_num_filters=config.bottleneck_num_filters,
        num_styles=config.num_styles),
    style_loss_func_factory_func=lambda: styleLoss.make_style_loss_function(
        styleLoss.StyleLossModelVGG(config.output_shape),
        config.output_shape,
        config.num_styles,
        config.with_depth_loss),
)

# call once to build model
log.info("Compiling model...")
style_transfer_training_model.loss_model.compile(run_eagerly=True)
log.info("Building model...")
style_transfer_training_model.loss_model(next(iter(validation_dataset)))
log.info(f"Loading weights from {checkpoint_path}")
# load_status = style_transfer_training_model.loss_model.load_weights(filepath=str(checkpoint_path))
checkpoint = tf.train.Checkpoint(style_transfer_training_model.inference)
load_status = checkpoint.restore(str(checkpoint_path))
load_status.assert_nontrivial_match()

log.info("Loading full dataset...")
preloaded_validation_dataset = list(progressify(validation_dataset, total=validation_dataset.num_samples))


for i, sample in enumerate(progressify(preloaded_validation_dataset, position=0, desc="Sample")):
    if permutation_importance_data.num_samples > i:
        continue

    matched_samples = list(preloaded_validation_dataset)
    del matched_samples[i]
    baseline_losses = style_transfer_training_model.loss_model(sample)

    for loss, loss_value in baseline_losses.items():
        prev_loss_value = permutation_importance_data.baseline_losses.setdefault(loss, 0)
        permutation_importance_data.baseline_losses[loss] = prev_loss_value + loss_value

    for j, matched_sample in enumerate(progressify(matched_samples, leave=False, position=1, desc="Permutation")):
        component_lower_bound = 0
        for channel, num_components in progressify(config.channels, leave=False, position=2, desc="Channel"):
            component_upper_bound = component_lower_bound + num_components
            permuted_sample_x, permuted_sample_y = sample
            permuted_sample_content = permuted_sample_x['content'].numpy()
            permuted_sample_content[..., component_lower_bound:component_upper_bound] = \
                matched_sample[0]['content'].numpy()[..., component_lower_bound:component_upper_bound]
            permuted_sample_x['content'] = tf.convert_to_tensor(permuted_sample_content)

            losses = style_transfer_training_model.loss_model((permuted_sample_x, permuted_sample_y))
            for loss, loss_value in losses.items():
                permutation_importance_data.channel_contributions[loss][channel] += loss_value \
                                                                                    - baseline_losses[loss]
            component_lower_bound = component_upper_bound

    permutation_importance_data.num_samples_and_permutations = (i + 1) * (len(preloaded_validation_dataset) - 1)
    permutation_importance_data.num_samples = i + 1

    permutation_importance_data_cache_file = permutation_importance_data_cache_file_path.open(mode='wb')
    pickle.dump(permutation_importance_data, permutation_importance_data_cache_file)

log.info(
    f"Averaging the results for {permutation_importance_data.num_samples_and_permutations} samples and permutations")
for loss, channels in permutation_importance_data.channel_contributions.items():
    for channel, channel_value in channels.items():
        permutation_importance_data.channel_contributions[loss][channel] = (
                channel_value / permutation_importance_data.num_samples_and_permutations).numpy().item()

for loss, loss_value in permutation_importance_data.baseline_losses.items():
    permutation_importance_data.baseline_losses[loss] = (
                loss_value / permutation_importance_data.num_samples).numpy().item()

import pandas as pd

losses_summary = {
    loss.replace("_", " "): {channel: contribution / permutation_importance_data.baseline_losses[loss] * 100 for channel, contribution in
           channels.items()} for loss, channels in permutation_importance_data.channel_contributions.items()}

print(pd.DataFrame.from_dict(permutation_importance_data.channel_contributions))
print(pd.DataFrame.from_dict({'baseline':permutation_importance_data.baseline_losses}, orient='index'))

importance_frame = pd.DataFrame\
    .from_dict(losses_summary, orient="columns")\
    .sort_values(by="loss",ascending=False)
# importance_frame['Baseline'] = permutation_importance_data.baseline_losses
print(importance_frame)
style = importance_frame.style
style = style.format('{:.02f}\\%')
print(style.to_latex())