File size: 6,785 Bytes
45099b6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
from typing import List

import numpy as np
import torch
from fastai.vision.all import *


# from backend.StampRemoval.util import *
__all__ = ["CustomUnetBlock", "CustomDynamicUnet", "UnetInference"]


class CustomUnetBlock(Module):
    """A quasi-UNet block, using `PixelShuffle_ICNR upsampling`."""

    @delegates(ConvLayer.__init__)
    def __init__(
        self,
        up_in_c,
        x_in_c,
        hook,
        final_div=True,
        blur=False,
        act_cls=defaults.activation,
        self_attention=False,
        init=nn.init.kaiming_normal_,
        norm_type=None,
        **kwargs,
    ):
        self.hook = hook
        self.shuf = PixelShuffle_ICNR(up_in_c, up_in_c // 2, blur=blur, act_cls=act_cls, norm_type=norm_type)
        self.bn = BatchNorm(x_in_c)
        ni = up_in_c // 2 + x_in_c
        #         nf = ni if final_div else ni//2
        nf = ni // 2 if final_div else ni // 4
        self.conv1 = ConvLayer(ni, nf, act_cls=act_cls, norm_type=norm_type, **kwargs)
        self.conv2 = ConvLayer(
            nf,
            nf,
            act_cls=act_cls,
            norm_type=norm_type,
            xtra=SelfAttention(nf) if self_attention else None,
            **kwargs,
        )
        self.relu = act_cls()
        apply_init(nn.Sequential(self.conv1, self.conv2), init)

    def forward(self, up_in):
        s = self.hook.stored
        up_out = self.shuf(up_in)
        ssh = s.shape[-2:]
        if ssh != up_out.shape[-2:]:
            up_out = F.interpolate(up_out, s.shape[-2:], mode="nearest")
        cat_x = self.relu(torch.cat([up_out, self.bn(s)], dim=1))
        return self.conv2(self.conv1(cat_x))


class CustomDynamicUnet(SequentialEx):
    """Create a U-Net from a given architecture."""

    def __init__(
        self,
        encoder,
        n_out,
        img_size,
        blur=False,
        blur_final=True,
        self_attention=False,
        y_range=None,
        last_cross=True,
        bottle=False,
        act_cls=defaults.activation,
        init=nn.init.kaiming_normal_,
        norm_type=None,
        **kwargs,
    ):
        imsize = img_size
        sizes = model_sizes(encoder, size=imsize)

        sz_chg_idxs = list(reversed(self._get_sz_change_idxs(sizes)))
        self.sfs = hook_outputs([encoder[i] for i in sz_chg_idxs], detach=False)
        x = dummy_eval(encoder, imsize).detach()

        ni = sizes[-1][1]

        middle_conv = nn.Sequential(
            ConvLayer(ni, ni, act_cls=act_cls, norm_type=norm_type, **kwargs),
            ConvLayer(ni, ni, act_cls=act_cls, norm_type=norm_type, **kwargs),
        ).eval()
        x = middle_conv(x)
        layers = [encoder, BatchNorm(ni), nn.ReLU(), middle_conv]

        for i, idx in enumerate(sz_chg_idxs):
            not_final = i != len(sz_chg_idxs) - 1
            up_in_c, x_in_c = int(x.shape[1]), int(sizes[idx][1])
            do_blur = blur and (not_final or blur_final)
            sa = self_attention and (i == len(sz_chg_idxs) - 3)
            unet_block = CustomUnetBlock(
                up_in_c,
                x_in_c,
                self.sfs[i],
                final_div=not_final,
                blur=do_blur,
                self_attention=sa,
                act_cls=act_cls,
                init=init,
                norm_type=norm_type,
                **kwargs,
            ).eval()
            layers.append(unet_block)
            x = unet_block(x)

        ni = x.shape[1]
        if imsize != sizes[0][-2:]:
            layers.append(PixelShuffle_ICNR(ni, act_cls=act_cls, norm_type=norm_type))
        layers.append(ResizeToOrig())
        if last_cross:
            layers.append(MergeLayer(dense=True))
            ni += in_channels(encoder)
            layers.append(
                ResBlock(
                    1,
                    ni,
                    ni // 2 if bottle else ni,
                    act_cls=act_cls,
                    norm_type=norm_type,
                    **kwargs,
                )
            )
        layers += [ConvLayer(ni, n_out, ks=1, act_cls=None, norm_type=norm_type, **kwargs)]
        apply_init(nn.Sequential(layers[3], layers[-2]), init)
        # apply_init(nn.Sequential(layers[2]), init)
        if y_range is not None:
            layers.append(SigmoidRange(*y_range))
        super().__init__(*layers)

    def _get_sz_change_idxs(self, sizes):
        "Get the indexes of the layers where the size of the activation changes."
        feature_szs = [size[-1] for size in sizes]
        sz_chg_idxs = list(np.where(np.array(feature_szs[:-1]) != np.array(feature_szs[1:]))[0])
        return sz_chg_idxs

    def __del__(self):
        if hasattr(self, "sfs"):
            self.sfs.remove()


class PerceptualLoss:
    pass


class UnetInference:
    def __init__(self, model_path):
        """Inference interface for unet model"""
        self.learn = load_learner(model_path)
        self.learn.model.eval()

    def __call__(self, image_array: str, bs: int = 1) -> List[np.ndarray]:
        """Perform forward pass and decode the prediction of Unet model

        Args:
            image_array (list): list of numpy array
            bs (int, optional): [batch size]. Defaults to 1.

        Returns:
            [list]: list of numpy array
        """
        if len(image_array) < 1:
            return []

        batches = self.__build_batches(image_array, bs=bs)
        outs = []
        with torch.no_grad():
            for b in batches:
                outs.append(self.learn.model(b))
                del b
        pil_images = self.__decode_prediction(outs)
        return pil_images

    def __decode_prediction(self, preds):
        out = []
        i2f = IntToFloatTensor()
        for pred in preds:
            img_np = i2f.decodes(pred.squeeze()).numpy()
            img_np = img_np.transpose(1, 2, 0)
            img_np = img_np.astype(np.uint8)
            out.append(img_np)
            # out.append(Image.fromarray(img_np))
            del img_np
        return out

    def __build_batches(self, image_array: list, bs=1):
        "Builds batches to skip `DataLoader` overhead"
        type_tfms = [PILImage.create]
        item_tfms = [ToTensor()]
        type_pipe = Pipeline(type_tfms)
        item_pipe = Pipeline(item_tfms)
        i2f = IntToFloatTensor()
        batches = []
        batch = []
        k = 0
        for i, im in enumerate(image_array):
            batch.append(item_pipe(type_pipe(im)))
            k += 1
            if i == len(image_array) - 1 or k == bs:
                # batches.append(torch.cat([norm(i2f(b.cuda())) for b in batch]))
                batches.append(torch.stack([i2f(b.cpu()) for b in batch], axis=0))
                batch = []
                k = 0
        return batches