NegiTurkey
commited on
Commit
•
b0637c4
1
Parent(s):
c2e0a3f
Update app.py
Browse files
app.py
CHANGED
@@ -6,13 +6,15 @@ import torch
|
|
6 |
from torchvision import transforms
|
7 |
import os
|
8 |
import zipfile
|
|
|
|
|
9 |
|
10 |
torch.set_float32_matmul_precision(["high", "highest"][0])
|
11 |
|
12 |
birefnet = AutoModelForImageSegmentation.from_pretrained(
|
13 |
"ZhengPeng7/BiRefNet", trust_remote_code=True
|
14 |
)
|
15 |
-
birefnet.to("
|
16 |
transform_image = transforms.Compose(
|
17 |
[
|
18 |
transforms.Resize((1024, 1024)),
|
@@ -25,7 +27,7 @@ def fn(image):
|
|
25 |
im = load_img(image, output_type="pil")
|
26 |
im = im.convert("RGB")
|
27 |
image_size = im.size
|
28 |
-
input_images = transform_image(im).unsqueeze(0).to("
|
29 |
|
30 |
with torch.no_grad():
|
31 |
preds = birefnet(input_images)[-1].sigmoid().cpu()
|
@@ -37,13 +39,16 @@ def fn(image):
|
|
37 |
output_file_path = os.path.join("output_images", "output_image_single.png")
|
38 |
im.save(output_file_path)
|
39 |
|
40 |
-
|
|
|
|
|
|
|
41 |
|
42 |
def fn_url(url):
|
43 |
im = load_img(url, output_type="pil")
|
44 |
im = im.convert("RGB")
|
45 |
image_size = im.size
|
46 |
-
input_images = transform_image(im).unsqueeze(0).to("
|
47 |
|
48 |
with torch.no_grad():
|
49 |
preds = birefnet(input_images)[-1].sigmoid().cpu()
|
@@ -55,7 +60,10 @@ def fn_url(url):
|
|
55 |
output_file_path = os.path.join("output_images", "output_image_url.png")
|
56 |
im.save(output_file_path)
|
57 |
|
58 |
-
|
|
|
|
|
|
|
59 |
|
60 |
def batch_fn(images):
|
61 |
output_paths = []
|
@@ -63,7 +71,7 @@ def batch_fn(images):
|
|
63 |
im = load_img(image_path, output_type="pil")
|
64 |
im = im.convert("RGB")
|
65 |
image_size = im.size
|
66 |
-
input_images = transform_image(im).unsqueeze(0).to("
|
67 |
|
68 |
with torch.no_grad():
|
69 |
preds = birefnet(input_images)[-1].sigmoid().cpu()
|
@@ -71,7 +79,7 @@ def batch_fn(images):
|
|
71 |
pred_pil = transforms.ToPILImage()(pred)
|
72 |
mask = pred_pil.resize(image_size)
|
73 |
|
74 |
-
im.putalpha(mask)
|
75 |
|
76 |
output_file_path = os.path.join("output_images", f"output_image_batch_{idx + 1}.png")
|
77 |
im.save(output_file_path)
|
@@ -95,10 +103,10 @@ chameleon = load_img("chameleon.jpg", output_type="pil")
|
|
95 |
url = "https://hips.hearstapps.com/hmg-prod/images/gettyimages-1229892983-square.jpg"
|
96 |
|
97 |
tab1 = gr.Interface(
|
98 |
-
fn, inputs=image, outputs=slider1, examples=[chameleon], api_name="image"
|
99 |
)
|
100 |
|
101 |
-
tab2 = gr.Interface(fn_url, inputs=text, outputs=slider2, examples=[url], api_name="text")
|
102 |
|
103 |
tab3 = gr.Interface(
|
104 |
batch_fn,
|
|
|
6 |
from torchvision import transforms
|
7 |
import os
|
8 |
import zipfile
|
9 |
+
import numpy as np
|
10 |
+
from PIL import Image
|
11 |
|
12 |
torch.set_float32_matmul_precision(["high", "highest"][0])
|
13 |
|
14 |
birefnet = AutoModelForImageSegmentation.from_pretrained(
|
15 |
"ZhengPeng7/BiRefNet", trust_remote_code=True
|
16 |
)
|
17 |
+
birefnet.to("cuda")
|
18 |
transform_image = transforms.Compose(
|
19 |
[
|
20 |
transforms.Resize((1024, 1024)),
|
|
|
27 |
im = load_img(image, output_type="pil")
|
28 |
im = im.convert("RGB")
|
29 |
image_size = im.size
|
30 |
+
input_images = transform_image(im).unsqueeze(0).to("cuda")
|
31 |
|
32 |
with torch.no_grad():
|
33 |
preds = birefnet(input_images)[-1].sigmoid().cpu()
|
|
|
39 |
output_file_path = os.path.join("output_images", "output_image_single.png")
|
40 |
im.save(output_file_path)
|
41 |
|
42 |
+
output_path = os.path.join("output_images", "output_image_processed.png")
|
43 |
+
im.save(output_path, "PNG")
|
44 |
+
|
45 |
+
return [im, mask], output_path
|
46 |
|
47 |
def fn_url(url):
|
48 |
im = load_img(url, output_type="pil")
|
49 |
im = im.convert("RGB")
|
50 |
image_size = im.size
|
51 |
+
input_images = transform_image(im).unsqueeze(0).to("cuda")
|
52 |
|
53 |
with torch.no_grad():
|
54 |
preds = birefnet(input_images)[-1].sigmoid().cpu()
|
|
|
60 |
output_file_path = os.path.join("output_images", "output_image_url.png")
|
61 |
im.save(output_file_path)
|
62 |
|
63 |
+
output_path = os.path.join("output_images", "output_image_url_processed.png")
|
64 |
+
im.save(output_path, "PNG")
|
65 |
+
|
66 |
+
return [im, mask], output_path
|
67 |
|
68 |
def batch_fn(images):
|
69 |
output_paths = []
|
|
|
71 |
im = load_img(image_path, output_type="pil")
|
72 |
im = im.convert("RGB")
|
73 |
image_size = im.size
|
74 |
+
input_images = transform_image(im).unsqueeze(0).to("cuda")
|
75 |
|
76 |
with torch.no_grad():
|
77 |
preds = birefnet(input_images)[-1].sigmoid().cpu()
|
|
|
79 |
pred_pil = transforms.ToPILImage()(pred)
|
80 |
mask = pred_pil.resize(image_size)
|
81 |
|
82 |
+
im.putalpha(mask)
|
83 |
|
84 |
output_file_path = os.path.join("output_images", f"output_image_batch_{idx + 1}.png")
|
85 |
im.save(output_file_path)
|
|
|
103 |
url = "https://hips.hearstapps.com/hmg-prod/images/gettyimages-1229892983-square.jpg"
|
104 |
|
105 |
tab1 = gr.Interface(
|
106 |
+
fn, inputs=image, outputs=[slider1, gr.File(label="PNG Output")], examples=[chameleon], api_name="image"
|
107 |
)
|
108 |
|
109 |
+
tab2 = gr.Interface(fn_url, inputs=text, outputs=[slider2, gr.File(label="PNG Output")], examples=[url], api_name="text")
|
110 |
|
111 |
tab3 = gr.Interface(
|
112 |
batch_fn,
|