Spaces:
Sleeping
Sleeping
ZeroGPU (#2)
Browse files- Use Gradio 4.x so it can work with ZeroGPU (a8d124d9c329f5d663cb7ce5f3ed6767c7655e19)
Co-authored-by: hysts <hysts@users.noreply.huggingface.co>
README.md
CHANGED
@@ -4,7 +4,7 @@ emoji: 🌖
|
|
4 |
colorFrom: red
|
5 |
colorTo: indigo
|
6 |
sdk: gradio
|
7 |
-
sdk_version:
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
license: apache-2.0
|
|
|
4 |
colorFrom: red
|
5 |
colorTo: indigo
|
6 |
sdk: gradio
|
7 |
+
sdk_version: 4.14.0
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
license: apache-2.0
|
app.py
CHANGED
@@ -2,6 +2,7 @@ import gradio as gr
|
|
2 |
import cv2
|
3 |
import numpy as np
|
4 |
from PIL import Image
|
|
|
5 |
import torch
|
6 |
import torch.nn.functional as F
|
7 |
from torchvision.transforms import Compose
|
@@ -20,14 +21,14 @@ css = """
|
|
20 |
#img-display-output {
|
21 |
max-height: 80vh;
|
22 |
}
|
23 |
-
|
24 |
"""
|
25 |
DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
|
26 |
model = DPT_DINOv2(encoder='vitl', features=256, out_channels=[256, 512, 1024, 1024]).to(DEVICE).eval()
|
27 |
model.load_state_dict(torch.load('checkpoints/depth_anything_vitl14.pth'))
|
28 |
|
29 |
title = "# Depth Anything"
|
30 |
-
description = """Official demo for **Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data**.
|
31 |
|
32 |
Please refer to our [paper](), [project page](https://depth-anything.github.io), or [github](https://github.com/LiheYoung/Depth-Anything) for more details."""
|
33 |
|
@@ -45,38 +46,44 @@ transform = Compose([
|
|
45 |
PrepareForNet(),
|
46 |
])
|
47 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
with gr.Blocks(css=css) as demo:
|
49 |
gr.Markdown(title)
|
50 |
gr.Markdown(description)
|
51 |
gr.Markdown("### Depth Prediction demo")
|
52 |
-
|
53 |
with gr.Row():
|
54 |
-
input_image = gr.Image(label="Input Image", type='numpy', elem_id='img-display-input')
|
55 |
depth_image = gr.Image(label="Depth Map", elem_id='img-display-output')
|
56 |
raw_file = gr.File(label="16-bit raw depth (can be considered as disparity)")
|
57 |
submit = gr.Button("Submit")
|
58 |
|
59 |
def on_submit(image):
|
60 |
h, w = image.shape[:2]
|
61 |
-
|
62 |
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) / 255.0
|
63 |
image = transform({'image': image})['image']
|
64 |
image = torch.from_numpy(image).unsqueeze(0).to(DEVICE)
|
65 |
-
|
66 |
-
|
67 |
-
depth = model(image)
|
68 |
depth = F.interpolate(depth[None], (h, w), mode='bilinear', align_corners=False)[0, 0]
|
69 |
-
|
70 |
raw_depth = Image.fromarray(depth.cpu().numpy().astype('uint16'))
|
71 |
tmp = tempfile.NamedTemporaryFile(suffix='.png', delete=False)
|
72 |
raw_depth.save(tmp.name)
|
73 |
-
|
74 |
depth = (depth - depth.min()) / (depth.max() - depth.min()) * 255.0
|
75 |
depth = depth.cpu().numpy().astype(np.uint8)
|
76 |
colored_depth = cv2.applyColorMap(depth, cv2.COLORMAP_INFERNO)[:, :, ::-1]
|
77 |
-
|
78 |
return [colored_depth, tmp.name]
|
79 |
-
|
80 |
submit.click(on_submit, inputs=[input_image], outputs=[depth_image, raw_file])
|
81 |
examples = gr.Examples(examples=["examples/flower.png", "examples/roller_coaster.png", "examples/hall.png", "examples/car.png", "examples/person.png"],
|
82 |
inputs=[input_image])
|
|
|
2 |
import cv2
|
3 |
import numpy as np
|
4 |
from PIL import Image
|
5 |
+
import spaces
|
6 |
import torch
|
7 |
import torch.nn.functional as F
|
8 |
from torchvision.transforms import Compose
|
|
|
21 |
#img-display-output {
|
22 |
max-height: 80vh;
|
23 |
}
|
24 |
+
|
25 |
"""
|
26 |
DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
|
27 |
model = DPT_DINOv2(encoder='vitl', features=256, out_channels=[256, 512, 1024, 1024]).to(DEVICE).eval()
|
28 |
model.load_state_dict(torch.load('checkpoints/depth_anything_vitl14.pth'))
|
29 |
|
30 |
title = "# Depth Anything"
|
31 |
+
description = """Official demo for **Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data**.
|
32 |
|
33 |
Please refer to our [paper](), [project page](https://depth-anything.github.io), or [github](https://github.com/LiheYoung/Depth-Anything) for more details."""
|
34 |
|
|
|
46 |
PrepareForNet(),
|
47 |
])
|
48 |
|
49 |
+
|
50 |
+
@spaces.GPU
|
51 |
+
@torch.no_grad()
|
52 |
+
def predict_depth(model, image):
|
53 |
+
return model(image)
|
54 |
+
|
55 |
+
|
56 |
with gr.Blocks(css=css) as demo:
|
57 |
gr.Markdown(title)
|
58 |
gr.Markdown(description)
|
59 |
gr.Markdown("### Depth Prediction demo")
|
60 |
+
|
61 |
with gr.Row():
|
62 |
+
input_image = gr.Image(label="Input Image", type='numpy', elem_id='img-display-input')
|
63 |
depth_image = gr.Image(label="Depth Map", elem_id='img-display-output')
|
64 |
raw_file = gr.File(label="16-bit raw depth (can be considered as disparity)")
|
65 |
submit = gr.Button("Submit")
|
66 |
|
67 |
def on_submit(image):
|
68 |
h, w = image.shape[:2]
|
69 |
+
|
70 |
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) / 255.0
|
71 |
image = transform({'image': image})['image']
|
72 |
image = torch.from_numpy(image).unsqueeze(0).to(DEVICE)
|
73 |
+
|
74 |
+
depth = predict_depth(model, image)
|
|
|
75 |
depth = F.interpolate(depth[None], (h, w), mode='bilinear', align_corners=False)[0, 0]
|
76 |
+
|
77 |
raw_depth = Image.fromarray(depth.cpu().numpy().astype('uint16'))
|
78 |
tmp = tempfile.NamedTemporaryFile(suffix='.png', delete=False)
|
79 |
raw_depth.save(tmp.name)
|
80 |
+
|
81 |
depth = (depth - depth.min()) / (depth.max() - depth.min()) * 255.0
|
82 |
depth = depth.cpu().numpy().astype(np.uint8)
|
83 |
colored_depth = cv2.applyColorMap(depth, cv2.COLORMAP_INFERNO)[:, :, ::-1]
|
84 |
+
|
85 |
return [colored_depth, tmp.name]
|
86 |
+
|
87 |
submit.click(on_submit, inputs=[input_image], outputs=[depth_image, raw_file])
|
88 |
examples = gr.Examples(examples=["examples/flower.png", "examples/roller_coaster.png", "examples/hall.png", "examples/car.png", "examples/person.png"],
|
89 |
inputs=[input_image])
|