Spaces:
Running
on
Zero
Running
on
Zero
xiank he
commited on
Commit
·
72cd992
1
Parent(s):
89a1e10
distill-any-depth
Browse files
app.py
CHANGED
@@ -1,32 +1,42 @@
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
from PIL import Image
|
4 |
-
import cv2
|
5 |
import numpy as np
|
6 |
from distillanydepth.modeling.archs.dam.dam import DepthAnything
|
7 |
-
from distillanydepth.utils.image_util import colorize_depth_maps
|
8 |
from distillanydepth.midas.transforms import Resize, NormalizeImage, PrepareForNet
|
9 |
from torchvision.transforms import Compose
|
10 |
-
import
|
|
|
|
|
11 |
|
12 |
-
# Helper function to load model
|
13 |
def load_model_by_name(arch_name, checkpoint_path, device):
|
|
|
14 |
if arch_name == 'depthanything':
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
|
|
|
|
|
|
|
|
19 |
else:
|
20 |
raise NotImplementedError(f"Unknown architecture: {arch_name}")
|
21 |
return model
|
22 |
|
23 |
-
# Image processing function
|
24 |
def process_image(image, model, device):
|
|
|
|
|
|
|
25 |
# Preprocess the image
|
26 |
image_np = np.array(image)[..., ::-1] / 255
|
|
|
27 |
transform = Compose([
|
28 |
-
Resize(
|
29 |
-
NormalizeImage(mean=[0.
|
30 |
PrepareForNet()
|
31 |
])
|
32 |
|
@@ -35,31 +45,74 @@ def process_image(image, model, device):
|
|
35 |
|
36 |
with torch.no_grad(): # Disable autograd since we don't need gradients on CPU
|
37 |
pred_disp, _ = model(image_tensor)
|
38 |
-
pred_disp_np = pred_disp.cpu().detach().numpy()[0, :, :, :].transpose(1, 2, 0)
|
39 |
-
pred_disp = (pred_disp_np - pred_disp_np.min()) / (pred_disp_np.max() - pred_disp_np.min())
|
40 |
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
# Colorize depth map
|
42 |
-
cmap = "Spectral_r"
|
43 |
-
depth_colored = colorize_depth_maps(pred_disp[None,
|
|
|
|
|
44 |
depth_colored = (depth_colored * 255).astype(np.uint8)
|
45 |
|
46 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
47 |
return depth_image
|
48 |
|
49 |
# Gradio interface function
|
50 |
def gradio_interface(image):
|
51 |
-
|
52 |
-
|
53 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
54 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
# Process image and return output
|
56 |
-
|
|
|
57 |
|
58 |
# Create Gradio interface
|
59 |
iface = gr.Interface(
|
60 |
fn=gradio_interface,
|
61 |
inputs=gr.Image(type="pil"), # Only image input, no mode selection
|
62 |
-
outputs=gr.Image(type="pil"),
|
63 |
title="Depth Estimation Demo",
|
64 |
description="Upload an image to see the depth estimation results."
|
65 |
)
|
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
from PIL import Image
|
|
|
4 |
import numpy as np
|
5 |
from distillanydepth.modeling.archs.dam.dam import DepthAnything
|
6 |
+
from distillanydepth.utils.image_util import chw2hwc, colorize_depth_maps
|
7 |
from distillanydepth.midas.transforms import Resize, NormalizeImage, PrepareForNet
|
8 |
from torchvision.transforms import Compose
|
9 |
+
import cv2
|
10 |
+
from huggingface_hub import hf_hub_download
|
11 |
+
from safetensors.torch import load_file # 导入 safetensors 库
|
12 |
|
13 |
+
# Helper function to load model from Hugging Face
|
14 |
def load_model_by_name(arch_name, checkpoint_path, device):
|
15 |
+
model = None
|
16 |
if arch_name == 'depthanything':
|
17 |
+
# 使用 safetensors 加载模型权重
|
18 |
+
model_weights = load_file(checkpoint_path) # safetensors 加载方式
|
19 |
+
|
20 |
+
# 初始化模型
|
21 |
+
model = DepthAnything(checkpoint_path=None).to(device)
|
22 |
+
model.load_state_dict(model_weights) # 将加载的权重应用到模型
|
23 |
+
|
24 |
+
model = model.to(device) # 确保模型在正确的设备上
|
25 |
else:
|
26 |
raise NotImplementedError(f"Unknown architecture: {arch_name}")
|
27 |
return model
|
28 |
|
29 |
+
# Image processing function
|
30 |
def process_image(image, model, device):
|
31 |
+
if model is None:
|
32 |
+
return None
|
33 |
+
|
34 |
# Preprocess the image
|
35 |
image_np = np.array(image)[..., ::-1] / 255
|
36 |
+
|
37 |
transform = Compose([
|
38 |
+
Resize(756, 756, resize_target=False, keep_aspect_ratio=True, ensure_multiple_of=14, resize_method='lower_bound', image_interpolation_method=cv2.INTER_CUBIC),
|
39 |
+
NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
|
40 |
PrepareForNet()
|
41 |
])
|
42 |
|
|
|
45 |
|
46 |
with torch.no_grad(): # Disable autograd since we don't need gradients on CPU
|
47 |
pred_disp, _ = model(image_tensor)
|
|
|
|
|
48 |
|
49 |
+
# Ensure the depth map is in the correct shape before colorization
|
50 |
+
pred_disp_np = pred_disp.cpu().detach().numpy()[0, 0, :, :] # Remove extra singleton dimensions
|
51 |
+
|
52 |
+
# Normalize depth map
|
53 |
+
pred_disp = (pred_disp_np - pred_disp_np.min()) / (pred_disp_np.max() - pred_disp_np.min())
|
54 |
+
|
55 |
# Colorize depth map
|
56 |
+
cmap = "Spectral_r"
|
57 |
+
depth_colored = colorize_depth_maps(pred_disp[None, ..., None], 0, 1, cmap=cmap).squeeze() # Ensure correct dimension
|
58 |
+
|
59 |
+
# Convert to uint8 for image display
|
60 |
depth_colored = (depth_colored * 255).astype(np.uint8)
|
61 |
|
62 |
+
# Convert to HWC format (height, width, channels)
|
63 |
+
depth_colored_hwc = chw2hwc(depth_colored)
|
64 |
+
|
65 |
+
# Resize to match the original image dimensions (height, width)
|
66 |
+
h, w = image_np.shape[:2]
|
67 |
+
depth_colored_hwc = cv2.resize(depth_colored_hwc, (w, h), cv2.INTER_LINEAR)
|
68 |
+
|
69 |
+
# Convert to a PIL image
|
70 |
+
depth_image = Image.fromarray(depth_colored_hwc)
|
71 |
return depth_image
|
72 |
|
73 |
# Gradio interface function
|
74 |
def gradio_interface(image):
|
75 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
76 |
+
|
77 |
+
model_kwargs = dict(
|
78 |
+
vitb=dict(
|
79 |
+
encoder='vitb',
|
80 |
+
features=128,
|
81 |
+
out_channels=[96, 192, 384, 768],
|
82 |
+
),
|
83 |
+
vitl=dict(
|
84 |
+
encoder="vitl",
|
85 |
+
features=256,
|
86 |
+
out_channels=[256, 512, 1024, 1024],
|
87 |
+
use_bn=False,
|
88 |
+
use_clstoken=False,
|
89 |
+
max_depth=150.0,
|
90 |
+
mode='disparity',
|
91 |
+
pretrain_type='dinov2',
|
92 |
+
del_mask_token=False
|
93 |
+
)
|
94 |
+
)
|
95 |
+
# Load model
|
96 |
+
model = DepthAnything(**model_kwargs['vitl']).to(device)
|
97 |
+
checkpoint_path = hf_hub_download(repo_id=f"xingyang1/Distill-Any-Depth", filename=f"large/model.safetensors", repo_type="model")
|
98 |
|
99 |
+
# 使用 safetensors 加载模型权重
|
100 |
+
model_weights = load_file(checkpoint_path) # safetensors 加载方式
|
101 |
+
model.load_state_dict(model_weights)
|
102 |
+
model = model.to(device) # 确保模型在正确的设备上
|
103 |
+
|
104 |
+
if model is None:
|
105 |
+
return None
|
106 |
+
|
107 |
# Process image and return output
|
108 |
+
depth_image = process_image(image, model, device)
|
109 |
+
return depth_image
|
110 |
|
111 |
# Create Gradio interface
|
112 |
iface = gr.Interface(
|
113 |
fn=gradio_interface,
|
114 |
inputs=gr.Image(type="pil"), # Only image input, no mode selection
|
115 |
+
outputs=gr.Image(type="pil"), # Only depth image output, no debug info
|
116 |
title="Depth Estimation Demo",
|
117 |
description="Upload an image to see the depth estimation results."
|
118 |
)
|