demo with model

.gitattributes

@@ -25,3 +25,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zstandard filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+best_model.ckpt filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,3 @@
+venv
+__pycache__
+test.png

README.md

@@ -10,3 +10,8 @@ pinned: false
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+
+```
+pip install -r requirements.txt
+python3.8 app.py
+```

app.py

@@ -1,20 +1,54 @@
 import gradio as gr
 from PIL import Image
-import torchvision
+from collections import OrderedDict
 import torch
+from models.model import GLPDepth
+from PIL import Image
+from torchvision import transforms
+import matplotlib.pyplot as plt
+from matplotlib.backends.backend_agg import FigureCanvasAgg
+import numpy as np
+
+
+
 
 # load model
+DEVICE='cpu'
+def load_mde_model(path):
+    model = GLPDepth(max_depth=700.0, is_train=False).to(DEVICE)
+    model_weight = torch.load(path, map_location=torch.device('cpu'))
+    model_weight = model_weight['model_state_dict']
+    if 'module' in next(iter(model_weight.items()))[0]:
+        model_weight = OrderedDict((k[7:], v) for k, v in model_weight.items())
+    model.load_state_dict(model_weight)
+    model.eval()
+    return model
+
+model = load_mde_model('best_model.ckpt')
+preprocess = transforms.Compose([
+    transforms.Resize((512, 512)),
+    transforms.ToTensor()
+]) 
 
 def predict(input_image):
     pil_image = Image.fromarray(input_image.astype('uint8'), 'RGB')
     # transform image to torch and do preprocessing
-    torch_image = torchvision.transforms.ToTensor()(pil_image)
+    torch_img = preprocess(pil_image).to(DEVICE).unsqueeze(0)
     # model predict
-    prediction = torch.rand(torch_image.shape)
+    with torch.no_grad():
+        output_patch = model(torch_img)
     # transform torch to image
-    predicted_pil_image = torchvision.transforms.ToPILImage()(prediction)
+    predicted_image = output_patch['pred_d'].squeeze().cpu().detach().numpy()
     # return correct image
-    return predicted_pil_image
+    fig, ax = plt.subplots()
+    im = ax.imshow(predicted_image, cmap='jet', vmin=0, vmax=np.max(predicted_image))
+    plt.colorbar(im, ax=ax)
+
+    fig.canvas.draw()
+    data = np.frombuffer(fig.canvas.tostring_rgb(), dtype=np.uint8)
+    data = data.reshape(fig.canvas.get_width_height()[::-1] + (3,))
+
+    return data
 
 iface = gr.Interface(
     fn=predict, 

best_model.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccef15e6acd9e19231d0093d365d4a14c68454a83ac49ba8b292ce5df9ca4d23
+size 735542869

models/mit.py

@@ -0,0 +1,437 @@
+# ---------------------------------------------------------------
+# Copyright (c) 2021, NVIDIA Corporation. All rights reserved.
+#
+# This work is licensed under the NVIDIA Source Code License
+# ---------------------------------------------------------------
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from functools import partial
+
+from timm.models.layers import DropPath, to_2tuple, trunc_normal_
+from timm.models.registry import register_model
+from timm.models.vision_transformer import _cfg
+# from mmseg.models.builder import BACKBONES
+from mmcv.runner import load_checkpoint
+import math
+
+
+class Mlp(nn.Module):
+    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.):
+        super().__init__()
+        out_features = out_features or in_features
+        hidden_features = hidden_features or in_features
+        self.fc1 = nn.Linear(in_features, hidden_features)
+        self.dwconv = DWConv(hidden_features)
+        self.act = act_layer()
+        self.fc2 = nn.Linear(hidden_features, out_features)
+        self.drop = nn.Dropout(drop)
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+        elif isinstance(m, nn.Conv2d):
+            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
+            fan_out //= m.groups
+            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
+            if m.bias is not None:
+                m.bias.data.zero_()
+
+    def forward(self, x, H, W):
+        x = self.fc1(x)
+        x = self.dwconv(x, H, W)
+        x = self.act(x)
+        x = self.drop(x)
+        x = self.fc2(x)
+        x = self.drop(x)
+        return x
+
+
+class Attention(nn.Module):
+    def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0., sr_ratio=1):
+        super().__init__()
+        assert dim % num_heads == 0, f"dim {dim} should be divided by num_heads {num_heads}."
+
+        self.dim = dim
+        self.num_heads = num_heads
+        head_dim = dim // num_heads
+        self.scale = qk_scale or head_dim ** -0.5
+
+        self.q = nn.Linear(dim, dim, bias=qkv_bias)
+        self.kv = nn.Linear(dim, dim * 2, bias=qkv_bias)
+        self.attn_drop = nn.Dropout(attn_drop)
+        self.proj = nn.Linear(dim, dim)
+        self.proj_drop = nn.Dropout(proj_drop)
+
+        self.sr_ratio = sr_ratio
+        if sr_ratio > 1:
+            self.sr = nn.Conv2d(
+                dim, dim, kernel_size=sr_ratio, stride=sr_ratio)
+            self.norm = nn.LayerNorm(dim)
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+        elif isinstance(m, nn.Conv2d):
+            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
+            fan_out //= m.groups
+            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
+            if m.bias is not None:
+                m.bias.data.zero_()
+
+    def forward(self, x, H, W):
+        B, N, C = x.shape
+        q = self.q(x).reshape(B, N, self.num_heads, C //
+                              self.num_heads).permute(0, 2, 1, 3)
+
+        if self.sr_ratio > 1:
+            x_ = x.permute(0, 2, 1).reshape(B, C, H, W)
+            x_ = self.sr(x_).reshape(B, C, -1).permute(0, 2, 1)
+            x_ = self.norm(x_)
+            kv = self.kv(x_).reshape(B, -1, 2, self.num_heads,
+                                     C // self.num_heads).permute(2, 0, 3, 1, 4)
+        else:
+            kv = self.kv(x).reshape(B, -1, 2, self.num_heads, C //
+                                    self.num_heads).permute(2, 0, 3, 1, 4)
+        k, v = kv[0], kv[1]
+
+        attn = (q @ k.transpose(-2, -1)) * self.scale
+        attn = attn.softmax(dim=-1)
+        attn = self.attn_drop(attn)
+
+        x = (attn @ v).transpose(1, 2).reshape(B, N, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+
+        return x
+
+
+class Block(nn.Module):
+
+    def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0.,
+                 drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, sr_ratio=1):
+        super().__init__()
+        self.norm1 = norm_layer(dim)
+        self.attn = Attention(
+            dim,
+            num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale,
+            attn_drop=attn_drop, proj_drop=drop, sr_ratio=sr_ratio)
+        # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
+        self.drop_path = DropPath(
+            drop_path) if drop_path > 0. else nn.Identity()
+        self.norm2 = norm_layer(dim)
+        mlp_hidden_dim = int(dim * mlp_ratio)
+        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim,
+                       act_layer=act_layer, drop=drop)
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+        elif isinstance(m, nn.Conv2d):
+            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
+            fan_out //= m.groups
+            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
+            if m.bias is not None:
+                m.bias.data.zero_()
+
+    def forward(self, x, H, W):
+        x = x + self.drop_path(self.attn(self.norm1(x), H, W))
+        x = x + self.drop_path(self.mlp(self.norm2(x), H, W))
+
+        return x
+
+
+class OverlapPatchEmbed(nn.Module):
+    """ Image to Patch Embedding
+    """
+
+    def __init__(self, img_size=224, patch_size=7, stride=4, in_chans=3, embed_dim=768):
+        super().__init__()
+        img_size = to_2tuple(img_size)
+        patch_size = to_2tuple(patch_size)
+
+        self.img_size = img_size
+        self.patch_size = patch_size
+        self.H, self.W = img_size[0] // patch_size[0], img_size[1] // patch_size[1]
+        self.num_patches = self.H * self.W
+        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=stride,
+                              padding=(patch_size[0] // 2, patch_size[1] // 2))
+        self.norm = nn.LayerNorm(embed_dim)
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+        elif isinstance(m, nn.Conv2d):
+            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
+            fan_out //= m.groups
+            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
+            if m.bias is not None:
+                m.bias.data.zero_()
+
+    def forward(self, x):
+        x = self.proj(x)
+        _, _, H, W = x.shape
+        x = x.flatten(2).transpose(1, 2)
+        x = self.norm(x)
+
+        return x, H, W
+
+
+class MixVisionTransformer(nn.Module):
+    def __init__(self, img_size=224, patch_size=16, in_chans=3, num_classes=1000, embed_dims=[64, 128, 256, 512],
+                 num_heads=[1, 2, 4, 8], mlp_ratios=[4, 4, 4, 4], qkv_bias=False, qk_scale=None, drop_rate=0.,
+                 attn_drop_rate=0., drop_path_rate=0., norm_layer=nn.LayerNorm,
+                 depths=[3, 4, 6, 3], sr_ratios=[8, 4, 2, 1]):
+        super().__init__()
+        self.num_classes = num_classes
+        self.depths = depths
+
+        # patch_embed
+        self.patch_embed1 = OverlapPatchEmbed(img_size=img_size, patch_size=7, stride=4, in_chans=in_chans,
+                                              embed_dim=embed_dims[0])
+        self.patch_embed2 = OverlapPatchEmbed(img_size=img_size // 4, patch_size=3, stride=2, in_chans=embed_dims[0],
+                                              embed_dim=embed_dims[1])
+        self.patch_embed3 = OverlapPatchEmbed(img_size=img_size // 8, patch_size=3, stride=2, in_chans=embed_dims[1],
+                                              embed_dim=embed_dims[2])
+        self.patch_embed4 = OverlapPatchEmbed(img_size=img_size // 16, patch_size=3, stride=2, in_chans=embed_dims[2],
+                                              embed_dim=embed_dims[3])
+
+        # transformer encoder
+        dpr = [x.item() for x in torch.linspace(0, drop_path_rate,
+                                                sum(depths))]  # stochastic depth decay rule
+        cur = 0
+        self.block1 = nn.ModuleList([Block(
+            dim=embed_dims[0], num_heads=num_heads[0], mlp_ratio=mlp_ratios[0], qkv_bias=qkv_bias, qk_scale=qk_scale,
+            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur +
+                                                                    i], norm_layer=norm_layer,
+            sr_ratio=sr_ratios[0])
+            for i in range(depths[0])])
+        self.norm1 = norm_layer(embed_dims[0])
+
+        cur += depths[0]
+        self.block2 = nn.ModuleList([Block(
+            dim=embed_dims[1], num_heads=num_heads[1], mlp_ratio=mlp_ratios[1], qkv_bias=qkv_bias, qk_scale=qk_scale,
+            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur +
+                                                                    i], norm_layer=norm_layer,
+            sr_ratio=sr_ratios[1])
+            for i in range(depths[1])])
+        self.norm2 = norm_layer(embed_dims[1])
+
+        cur += depths[1]
+        self.block3 = nn.ModuleList([Block(
+            dim=embed_dims[2], num_heads=num_heads[2], mlp_ratio=mlp_ratios[2], qkv_bias=qkv_bias, qk_scale=qk_scale,
+            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur +
+                                                                    i], norm_layer=norm_layer,
+            sr_ratio=sr_ratios[2])
+            for i in range(depths[2])])
+        self.norm3 = norm_layer(embed_dims[2])
+
+        cur += depths[2]
+        self.block4 = nn.ModuleList([Block(
+            dim=embed_dims[3], num_heads=num_heads[3], mlp_ratio=mlp_ratios[3], qkv_bias=qkv_bias, qk_scale=qk_scale,
+            drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[cur +
+                                                                    i], norm_layer=norm_layer,
+            sr_ratio=sr_ratios[3])
+            for i in range(depths[3])])
+        self.norm4 = norm_layer(embed_dims[3])
+
+        # classification head
+        # self.head = nn.Linear(embed_dims[3], num_classes) if num_classes > 0 else nn.Identity()
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+        elif isinstance(m, nn.Conv2d):
+            fan_out = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
+            fan_out //= m.groups
+            m.weight.data.normal_(0, math.sqrt(2.0 / fan_out))
+            if m.bias is not None:
+                m.bias.data.zero_()
+
+    def init_weights(self, pretrained=None):
+        if isinstance(pretrained, str):
+            load_checkpoint(self, pretrained, map_location='cpu',
+                            strict=False)
+
+    def reset_drop_path(self, drop_path_rate):
+        dpr = [x.item() for x in torch.linspace(
+            0, drop_path_rate, sum(self.depths))]
+        cur = 0
+        for i in range(self.depths[0]):
+            self.block1[i].drop_path.drop_prob = dpr[cur + i]
+
+        cur += self.depths[0]
+        for i in range(self.depths[1]):
+            self.block2[i].drop_path.drop_prob = dpr[cur + i]
+
+        cur += self.depths[1]
+        for i in range(self.depths[2]):
+            self.block3[i].drop_path.drop_prob = dpr[cur + i]
+
+        cur += self.depths[2]
+        for i in range(self.depths[3]):
+            self.block4[i].drop_path.drop_prob = dpr[cur + i]
+
+    def freeze_patch_emb(self):
+        self.patch_embed1.requires_grad = False
+
+    @torch.jit.ignore
+    def no_weight_decay(self):
+        # has pos_embed may be better
+        return {'pos_embed1', 'pos_embed2', 'pos_embed3', 'pos_embed4', 'cls_token'}
+
+    def get_classifier(self):
+        return self.head
+
+    def reset_classifier(self, num_classes, global_pool=''):
+        self.num_classes = num_classes
+        self.head = nn.Linear(
+            self.embed_dim, num_classes) if num_classes > 0 else nn.Identity()
+
+    def forward_features(self, x):
+        B = x.shape[0]
+        outs = []
+
+        # stage 1
+        x, H, W = self.patch_embed1(x)
+        for i, blk in enumerate(self.block1):
+            x = blk(x, H, W)
+        x = self.norm1(x)
+        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
+        outs.append(x)
+
+        # stage 2
+        x, H, W = self.patch_embed2(x)
+        for i, blk in enumerate(self.block2):
+            x = blk(x, H, W)
+        x = self.norm2(x)
+        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
+        outs.append(x)
+
+        # stage 3
+        x, H, W = self.patch_embed3(x)
+        for i, blk in enumerate(self.block3):
+            x = blk(x, H, W)
+        x = self.norm3(x)
+        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
+        outs.append(x)
+
+        # stage 4
+        x, H, W = self.patch_embed4(x)
+        for i, blk in enumerate(self.block4):
+            x = blk(x, H, W)
+        x = self.norm4(x)
+        x = x.reshape(B, H, W, -1).permute(0, 3, 1, 2).contiguous()
+        outs.append(x)
+
+        return outs
+
+    def forward(self, x):
+        x = self.forward_features(x)
+        # x = self.head(x)
+
+        return x
+
+
+class DWConv(nn.Module):
+    def __init__(self, dim=768):
+        super(DWConv, self).__init__()
+        self.dwconv = nn.Conv2d(dim, dim, 3, 1, 1, bias=True, groups=dim)
+
+    def forward(self, x, H, W):
+        B, N, C = x.shape
+        x = x.transpose(1, 2).view(B, C, H, W)
+        x = self.dwconv(x)
+        x = x.flatten(2).transpose(1, 2)
+
+        return x
+
+
+class mit_b0(MixVisionTransformer):
+    def __init__(self, **kwargs):
+        super(mit_b0, self).__init__(
+            patch_size=4, embed_dims=[32, 64, 160, 256], num_heads=[1, 2, 5, 8], mlp_ratios=[4, 4, 4, 4],
+            qkv_bias=True, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[2, 2, 2, 2], sr_ratios=[8, 4, 2, 1],
+            drop_rate=0.0, drop_path_rate=0.1)
+
+
+class mit_b1(MixVisionTransformer):
+    def __init__(self, **kwargs):
+        super(mit_b1, self).__init__(
+            patch_size=4, embed_dims=[64, 128, 320, 512], num_heads=[1, 2, 5, 8], mlp_ratios=[4, 4, 4, 4],
+            qkv_bias=True, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[2, 2, 2, 2], sr_ratios=[8, 4, 2, 1],
+            drop_rate=0.0, drop_path_rate=0.1)
+
+
+class mit_b2(MixVisionTransformer):
+    def __init__(self, **kwargs):
+        super(mit_b2, self).__init__(
+            patch_size=4, embed_dims=[64, 128, 320, 512], num_heads=[1, 2, 5, 8], mlp_ratios=[4, 4, 4, 4],
+            qkv_bias=True, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[3, 4, 6, 3], sr_ratios=[8, 4, 2, 1],
+            drop_rate=0.0, drop_path_rate=0.1)
+
+
+class mit_b3(MixVisionTransformer):
+    def __init__(self, **kwargs):
+        super(mit_b3, self).__init__(
+            patch_size=4, embed_dims=[64, 128, 320, 512], num_heads=[1, 2, 5, 8], mlp_ratios=[4, 4, 4, 4],
+            qkv_bias=True, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[3, 4, 18, 3], sr_ratios=[8, 4, 2, 1],
+            drop_rate=0.0, drop_path_rate=0.1)
+
+
+class mit_b4(MixVisionTransformer):
+    def __init__(self, **kwargs):
+        super(mit_b4, self).__init__(
+            patch_size=4, embed_dims=[64, 128, 320, 512], num_heads=[1, 2, 5, 8], mlp_ratios=[4, 4, 4, 4],
+            qkv_bias=True, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[3, 8, 27, 3], sr_ratios=[8, 4, 2, 1],
+            drop_rate=0.0, drop_path_rate=0.1)
+
+
+class mit_b5(MixVisionTransformer):
+    def __init__(self, **kwargs):
+        super(mit_b5, self).__init__(
+            patch_size=4, embed_dims=[64, 128, 320, 512], num_heads=[1, 2, 5, 8], mlp_ratios=[4, 4, 4, 4],
+            qkv_bias=True, norm_layer=partial(nn.LayerNorm, eps=1e-6), depths=[3, 6, 40, 3], sr_ratios=[8, 4, 2, 1],
+            drop_rate=0.0, drop_path_rate=0.1)
+
+
+if __name__ == "__main__":
+    import pdb
+
+    model = mit_b5()
+    pdb.set_trace()

models/model.py

@@ -0,0 +1,112 @@
+import torch
+import torch.nn as nn
+
+from mmcv.runner import load_checkpoint
+from models.mit import mit_b4
+
+
+class GLPDepth(nn.Module):
+    def __init__(self, max_depth=10.0, is_train=False):
+        super().__init__()
+        self.max_depth = max_depth
+
+        self.encoder = mit_b4()
+        if is_train:
+            ckpt_path = './models/weights/mit_b4.pth'
+            try:
+                load_checkpoint(self.encoder, ckpt_path, logger=None)
+            except:
+                import gdown
+                print("Download pre-trained encoder weights...")
+                id = '1BUtU42moYrOFbsMCE-LTTkUE-mrWnfG2'
+                url = 'https://drive.google.com/uc?id=' + id
+                output = './models/weights/mit_b4.pth'
+                gdown.download(url, output, quiet=False)
+
+        channels_in = [512, 320, 128]
+        channels_out = 64
+
+        self.decoder = Decoder(channels_in, channels_out)
+
+        self.last_layer_depth = nn.Sequential(
+            nn.Conv2d(channels_out, channels_out, kernel_size=3, stride=1, padding=1),
+            nn.ReLU(inplace=False),
+            nn.Conv2d(channels_out, 1, kernel_size=3, stride=1, padding=1))
+
+    def forward(self, x):
+        conv1, conv2, conv3, conv4 = self.encoder(x)
+        out = self.decoder(conv1, conv2, conv3, conv4)
+        out_depth = self.last_layer_depth(out)
+        out_depth = torch.sigmoid(out_depth) * self.max_depth
+
+        return {'pred_d': out_depth}
+
+
+class Decoder(nn.Module):
+    def __init__(self, in_channels, out_channels):
+        super().__init__()
+
+        self.bot_conv = nn.Conv2d(
+            in_channels=in_channels[0], out_channels=out_channels, kernel_size=1)
+        self.skip_conv1 = nn.Conv2d(
+            in_channels=in_channels[1], out_channels=out_channels, kernel_size=1)
+        self.skip_conv2 = nn.Conv2d(
+            in_channels=in_channels[2], out_channels=out_channels, kernel_size=1)
+
+        self.up = nn.Upsample(scale_factor=2, mode='bilinear', align_corners=False)
+
+        self.fusion1 = SelectiveFeatureFusion(out_channels)
+        self.fusion2 = SelectiveFeatureFusion(out_channels)
+        self.fusion3 = SelectiveFeatureFusion(out_channels)
+
+    def forward(self, x_1, x_2, x_3, x_4):
+        x_4_ = self.bot_conv(x_4)
+        out = self.up(x_4_)
+
+        x_3_ = self.skip_conv1(x_3)
+        out = self.fusion1(x_3_, out)
+        out = self.up(out)
+
+        x_2_ = self.skip_conv2(x_2)
+        out = self.fusion2(x_2_, out)
+        out = self.up(out)
+
+        out = self.fusion3(x_1, out)
+        out = self.up(out)
+        out = self.up(out)
+
+        return out
+
+
+class SelectiveFeatureFusion(nn.Module):
+    def __init__(self, in_channel=64):
+        super().__init__()
+
+        self.conv1 = nn.Sequential(
+            nn.Conv2d(in_channels=int(in_channel * 2),
+                      out_channels=in_channel, kernel_size=3, stride=1, padding=1),
+            nn.BatchNorm2d(in_channel),
+            nn.ReLU())
+
+        self.conv2 = nn.Sequential(
+            nn.Conv2d(in_channels=in_channel,
+                      out_channels=int(in_channel / 2), kernel_size=3, stride=1, padding=1),
+            nn.BatchNorm2d(int(in_channel / 2)),
+            nn.ReLU())
+
+        self.conv3 = nn.Conv2d(in_channels=int(in_channel / 2),
+                               out_channels=2, kernel_size=3, stride=1, padding=1)
+
+        self.sigmoid = nn.Sigmoid()
+
+    def forward(self, x_local, x_global):
+        x = torch.cat((x_local, x_global), dim=1)
+        x = self.conv1(x)
+        x = self.conv2(x)
+        x = self.conv3(x)
+        attn = self.sigmoid(x)
+
+        out = x_local * attn[:, 0, :, :].unsqueeze(1) + \
+              x_global * attn[:, 1, :, :].unsqueeze(1)
+
+        return out

requirements.txt

@@ -1,3 +1,5 @@
 gradio
 torch
 torchvision
+mmcv==1.4.3
+timm==0.5.4

test.py

@@ -0,0 +1,36 @@
+from collections import OrderedDict
+import torch
+from models.model import GLPDepth
+from PIL import Image
+from torchvision import transforms
+import matplotlib.pyplot as plt
+import numpy as np
+
+DEVICE='cpu'
+
+def load_mde_model(path):
+    model = GLPDepth(max_depth=700.0, is_train=False).to(DEVICE)
+    model_weight = torch.load(path, map_location=torch.device('cpu'))
+    model_weight = model_weight['model_state_dict']
+    if 'module' in next(iter(model_weight.items()))[0]:
+        model_weight = OrderedDict((k[7:], v) for k, v in model_weight.items())
+    model.load_state_dict(model_weight)
+    model.eval()
+    return model
+
+model = load_mde_model('best_model.ckpt')
+preprocess = transforms.Compose([
+    transforms.Resize((512, 512)),
+    transforms.ToTensor()
+]) 
+
+input_img = Image.open('demo_imgs/fake.jpg')
+torch_img = preprocess(input_img).to(DEVICE).unsqueeze(0)
+with torch.no_grad():
+    output_patch = model(torch_img)
+output_patch = output_patch['pred_d'].squeeze().cpu().detach().numpy()
+print(output_patch.shape)
+
+plt.imshow(output_patch, cmap='jet', vmin=0, vmax=np.max(output_patch))
+plt.colorbar()
+plt.savefig('test.png')