Spaces:
Running
on
Zero
Running
on
Zero
Upload app.py
Browse files
app.py
CHANGED
@@ -3,19 +3,24 @@ import numpy as np
|
|
3 |
import torch
|
4 |
from PIL import Image
|
5 |
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
from DAI.pipeline_all import DAIPipeline
|
7 |
from DAI.controlnetvae import ControlNetVAEModel
|
8 |
from DAI.decoder import CustomAutoencoderKL
|
9 |
from diffusers import AutoencoderKL, UNet2DConditionModel
|
10 |
from transformers import CLIPTextModel, AutoTokenizer
|
11 |
|
12 |
-
# Initialize device and model paths
|
13 |
-
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
14 |
-
weight_dtype = torch.float32
|
15 |
pretrained_model_name_or_path = "sjtu-deepvision/dereflection-any-image-v0"
|
16 |
pretrained_model_name_or_path2 = "stabilityai/stable-diffusion-2-1"
|
17 |
|
18 |
-
#
|
19 |
controlnet = ControlNetVAEModel.from_pretrained(pretrained_model_name_or_path, subfolder="controlnet", torch_dtype=weight_dtype).to(device)
|
20 |
unet = UNet2DConditionModel.from_pretrained(pretrained_model_name_or_path, subfolder="unet", torch_dtype=weight_dtype).to(device)
|
21 |
vae_2 = CustomAutoencoderKL.from_pretrained(pretrained_model_name_or_path, subfolder="vae_2", torch_dtype=weight_dtype).to(device)
|
@@ -23,7 +28,7 @@ vae = AutoencoderKL.from_pretrained(pretrained_model_name_or_path2, subfolder="v
|
|
23 |
text_encoder = CLIPTextModel.from_pretrained(pretrained_model_name_or_path2, subfolder="text_encoder").to(device)
|
24 |
tokenizer = AutoTokenizer.from_pretrained(pretrained_model_name_or_path2, subfolder="tokenizer", use_fast=False)
|
25 |
|
26 |
-
#
|
27 |
pipe = DAIPipeline(
|
28 |
vae=vae,
|
29 |
text_encoder=text_encoder,
|
@@ -36,12 +41,13 @@ pipe = DAIPipeline(
|
|
36 |
t_start=0,
|
37 |
).to(device)
|
38 |
|
39 |
-
#
|
|
|
40 |
def process_image(input_image):
|
41 |
-
#
|
42 |
input_image = Image.fromarray(input_image)
|
43 |
|
44 |
-
#
|
45 |
pipe_out = pipe(
|
46 |
image=input_image,
|
47 |
prompt="remove glass reflection",
|
@@ -49,16 +55,17 @@ def process_image(input_image):
|
|
49 |
processing_resolution=None,
|
50 |
)
|
51 |
|
52 |
-
#
|
53 |
processed_frame = (pipe_out.prediction.clip(-1, 1) + 1) / 2
|
54 |
processed_frame = (processed_frame[0] * 255).astype(np.uint8)
|
55 |
processed_frame = Image.fromarray(processed_frame)
|
56 |
|
57 |
-
|
|
|
58 |
|
59 |
-
# Gradio
|
60 |
def create_gradio_interface():
|
61 |
-
#
|
62 |
example_images = [
|
63 |
os.path.join("files", "image", f"{i}.png") for i in range(1, 9)
|
64 |
]
|
@@ -70,27 +77,33 @@ def create_gradio_interface():
|
|
70 |
input_image = gr.Image(label="Input Image", type="numpy")
|
71 |
submit_btn = gr.Button("Remove Reflection", variant="primary")
|
72 |
with gr.Column():
|
73 |
-
|
|
|
|
|
|
|
|
|
|
|
74 |
|
75 |
-
#
|
76 |
gr.Examples(
|
77 |
examples=example_images,
|
78 |
inputs=input_image,
|
79 |
-
outputs=
|
80 |
fn=process_image,
|
81 |
-
cache_examples=False, #
|
82 |
label="Example Images",
|
83 |
)
|
84 |
|
|
|
85 |
submit_btn.click(
|
86 |
fn=process_image,
|
87 |
inputs=input_image,
|
88 |
-
outputs=
|
89 |
)
|
90 |
|
91 |
return demo
|
92 |
|
93 |
-
#
|
94 |
def main():
|
95 |
demo = create_gradio_interface()
|
96 |
demo.launch(server_name="0.0.0.0", server_port=7860)
|
|
|
3 |
import torch
|
4 |
from PIL import Image
|
5 |
import gradio as gr
|
6 |
+
from gradio_imageslider import ImageSlider
|
7 |
+
import spaces # 必须放在最前面,确保 ZeroGPU 初始化
|
8 |
+
|
9 |
+
# 延迟 CUDA 初始化
|
10 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
11 |
+
weight_dtype = torch.float32
|
12 |
+
|
13 |
+
# 加载模型组件
|
14 |
from DAI.pipeline_all import DAIPipeline
|
15 |
from DAI.controlnetvae import ControlNetVAEModel
|
16 |
from DAI.decoder import CustomAutoencoderKL
|
17 |
from diffusers import AutoencoderKL, UNet2DConditionModel
|
18 |
from transformers import CLIPTextModel, AutoTokenizer
|
19 |
|
|
|
|
|
|
|
20 |
pretrained_model_name_or_path = "sjtu-deepvision/dereflection-any-image-v0"
|
21 |
pretrained_model_name_or_path2 = "stabilityai/stable-diffusion-2-1"
|
22 |
|
23 |
+
# 加载模型
|
24 |
controlnet = ControlNetVAEModel.from_pretrained(pretrained_model_name_or_path, subfolder="controlnet", torch_dtype=weight_dtype).to(device)
|
25 |
unet = UNet2DConditionModel.from_pretrained(pretrained_model_name_or_path, subfolder="unet", torch_dtype=weight_dtype).to(device)
|
26 |
vae_2 = CustomAutoencoderKL.from_pretrained(pretrained_model_name_or_path, subfolder="vae_2", torch_dtype=weight_dtype).to(device)
|
|
|
28 |
text_encoder = CLIPTextModel.from_pretrained(pretrained_model_name_or_path2, subfolder="text_encoder").to(device)
|
29 |
tokenizer = AutoTokenizer.from_pretrained(pretrained_model_name_or_path2, subfolder="tokenizer", use_fast=False)
|
30 |
|
31 |
+
# 创建推理管道
|
32 |
pipe = DAIPipeline(
|
33 |
vae=vae,
|
34 |
text_encoder=text_encoder,
|
|
|
41 |
t_start=0,
|
42 |
).to(device)
|
43 |
|
44 |
+
# 使用 spaces.GPU 包装推理函数
|
45 |
+
@spaces.GPU
|
46 |
def process_image(input_image):
|
47 |
+
# 将 Gradio 输入转换为 PIL 图像
|
48 |
input_image = Image.fromarray(input_image)
|
49 |
|
50 |
+
# 处理图像
|
51 |
pipe_out = pipe(
|
52 |
image=input_image,
|
53 |
prompt="remove glass reflection",
|
|
|
55 |
processing_resolution=None,
|
56 |
)
|
57 |
|
58 |
+
# 将输出转换为图像
|
59 |
processed_frame = (pipe_out.prediction.clip(-1, 1) + 1) / 2
|
60 |
processed_frame = (processed_frame[0] * 255).astype(np.uint8)
|
61 |
processed_frame = Image.fromarray(processed_frame)
|
62 |
|
63 |
+
# 返回输入图像和处理后的图像
|
64 |
+
return input_image, processed_frame
|
65 |
|
66 |
+
# 创建 Gradio 界面
|
67 |
def create_gradio_interface():
|
68 |
+
# 示例图像
|
69 |
example_images = [
|
70 |
os.path.join("files", "image", f"{i}.png") for i in range(1, 9)
|
71 |
]
|
|
|
77 |
input_image = gr.Image(label="Input Image", type="numpy")
|
78 |
submit_btn = gr.Button("Remove Reflection", variant="primary")
|
79 |
with gr.Column():
|
80 |
+
# 使用 ImageSlider 显示前后对比
|
81 |
+
output_slider = ImageSlider(
|
82 |
+
label="Before & After",
|
83 |
+
show_download_button=True,
|
84 |
+
show_share_button=True,
|
85 |
+
)
|
86 |
|
87 |
+
# 添加示例
|
88 |
gr.Examples(
|
89 |
examples=example_images,
|
90 |
inputs=input_image,
|
91 |
+
outputs=output_slider,
|
92 |
fn=process_image,
|
93 |
+
cache_examples=False, # 缓存结果以加快加载速度
|
94 |
label="Example Images",
|
95 |
)
|
96 |
|
97 |
+
# 绑定按钮点击事件
|
98 |
submit_btn.click(
|
99 |
fn=process_image,
|
100 |
inputs=input_image,
|
101 |
+
outputs=output_slider,
|
102 |
)
|
103 |
|
104 |
return demo
|
105 |
|
106 |
+
# 主函数
|
107 |
def main():
|
108 |
demo = create_gradio_interface()
|
109 |
demo.launch(server_name="0.0.0.0", server_port=7860)
|