File size: 6,673 Bytes
c84e839
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import argparse
import torch
import torch.nn as nn
import torch.nn.functional as F
from huggingface_hub import PyTorchModelHubMixin, hf_hub_download

from depth_anything.blocks import FeatureFusionBlock, _make_scratch


def _make_fusion_block(features, use_bn, size = None):
    return FeatureFusionBlock(
        features,
        nn.ReLU(False),
        deconv=False,
        bn=use_bn,
        expand=False,
        align_corners=True,
        size=size,
    )


class DPTHead(nn.Module):
    def __init__(self, nclass, in_channels, features=256, use_bn=False, out_channels=[256, 512, 1024, 1024], use_clstoken=False):
        super(DPTHead, self).__init__()
        
        self.nclass = nclass
        self.use_clstoken = use_clstoken
        
        self.projects = nn.ModuleList([
            nn.Conv2d(
                in_channels=in_channels,
                out_channels=out_channel,
                kernel_size=1,
                stride=1,
                padding=0,
            ) for out_channel in out_channels
        ])
        
        self.resize_layers = nn.ModuleList([
            nn.ConvTranspose2d(
                in_channels=out_channels[0],
                out_channels=out_channels[0],
                kernel_size=4,
                stride=4,
                padding=0),
            nn.ConvTranspose2d(
                in_channels=out_channels[1],
                out_channels=out_channels[1],
                kernel_size=2,
                stride=2,
                padding=0),
            nn.Identity(),
            nn.Conv2d(
                in_channels=out_channels[3],
                out_channels=out_channels[3],
                kernel_size=3,
                stride=2,
                padding=1)
        ])
        
        if use_clstoken:
            self.readout_projects = nn.ModuleList()
            for _ in range(len(self.projects)):
                self.readout_projects.append(
                    nn.Sequential(
                        nn.Linear(2 * in_channels, in_channels),
                        nn.GELU()))
        
        self.scratch = _make_scratch(
            out_channels,
            features,
            groups=1,
            expand=False,
        )

        self.scratch.stem_transpose = None
        
        self.scratch.refinenet1 = _make_fusion_block(features, use_bn)
        self.scratch.refinenet2 = _make_fusion_block(features, use_bn)
        self.scratch.refinenet3 = _make_fusion_block(features, use_bn)
        self.scratch.refinenet4 = _make_fusion_block(features, use_bn)

        head_features_1 = features
        head_features_2 = 32
        
        if nclass > 1:
            self.scratch.output_conv = nn.Sequential(
                nn.Conv2d(head_features_1, head_features_1, kernel_size=3, stride=1, padding=1),
                nn.ReLU(True),
                nn.Conv2d(head_features_1, nclass, kernel_size=1, stride=1, padding=0),
            )
        else:
            self.scratch.output_conv1 = nn.Conv2d(head_features_1, head_features_1 // 2, kernel_size=3, stride=1, padding=1)
            
            self.scratch.output_conv2 = nn.Sequential(
                nn.Conv2d(head_features_1 // 2, head_features_2, kernel_size=3, stride=1, padding=1),
                nn.ReLU(True),
                nn.Conv2d(head_features_2, 1, kernel_size=1, stride=1, padding=0),
                nn.ReLU(True),
                nn.Identity(),
            )
            
    def forward(self, out_features, patch_h, patch_w):
        out = []
        for i, x in enumerate(out_features):
            if self.use_clstoken:
                x, cls_token = x[0], x[1]
                readout = cls_token.unsqueeze(1).expand_as(x)
                x = self.readout_projects[i](torch.cat((x, readout), -1))
            else:
                x = x[0]
            
            x = x.permute(0, 2, 1).reshape((x.shape[0], x.shape[-1], patch_h, patch_w))
            
            x = self.projects[i](x)
            x = self.resize_layers[i](x)
            
            out.append(x)
        
        layer_1, layer_2, layer_3, layer_4 = out
        
        layer_1_rn = self.scratch.layer1_rn(layer_1)
        layer_2_rn = self.scratch.layer2_rn(layer_2)
        layer_3_rn = self.scratch.layer3_rn(layer_3)
        layer_4_rn = self.scratch.layer4_rn(layer_4)
        
        path_4 = self.scratch.refinenet4(layer_4_rn, size=layer_3_rn.shape[2:])
        path_3 = self.scratch.refinenet3(path_4, layer_3_rn, size=layer_2_rn.shape[2:])
        path_2 = self.scratch.refinenet2(path_3, layer_2_rn, size=layer_1_rn.shape[2:])
        path_1 = self.scratch.refinenet1(path_2, layer_1_rn)
        
        out = self.scratch.output_conv1(path_1)
        out = F.interpolate(out, (int(patch_h * 14), int(patch_w * 14)), mode="bilinear", align_corners=True)
        out = self.scratch.output_conv2(out)
        
        return out
        
        
class DPT_DINOv2(nn.Module):
    def __init__(self, encoder='vitl', features=256, out_channels=[256, 512, 1024, 1024], use_bn=False, use_clstoken=False, localhub=True):
        super(DPT_DINOv2, self).__init__()
        
        assert encoder in ['vits', 'vitb', 'vitl']
        
        # in case the Internet connection is not stable, please load the DINOv2 locally
        if localhub:
            self.pretrained = torch.hub.load('torchhub/facebookresearch_dinov2_main', 'dinov2_{:}14'.format(encoder), source='local', pretrained=False)
        else:
            self.pretrained = torch.hub.load('facebookresearch/dinov2', 'dinov2_{:}14'.format(encoder))
        
        dim = self.pretrained.blocks[0].attn.qkv.in_features
        
        self.depth_head = DPTHead(1, dim, features, use_bn, out_channels=out_channels, use_clstoken=use_clstoken)
        
    def forward(self, x):
        h, w = x.shape[-2:]
        
        features = self.pretrained.get_intermediate_layers(x, 4, return_class_token=True)
        
        patch_h, patch_w = h // 14, w // 14

        depth = self.depth_head(features, patch_h, patch_w)
        depth = F.interpolate(depth, size=(h, w), mode="bilinear", align_corners=True)
        depth = F.relu(depth)

        return depth.squeeze(1)


class DepthAnything(DPT_DINOv2, PyTorchModelHubMixin):
    def __init__(self, config):
        super().__init__(**config)


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument(
        "--encoder",
        default="vits",
        type=str,
        choices=["vits", "vitb", "vitl"],
    )
    args = parser.parse_args()
    
    model = DepthAnything.from_pretrained("LiheYoung/depth_anything_{:}14".format(args.encoder))
    
    print(model)