Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
|
@@ -20,25 +20,19 @@ from transformers import GroundingDinoForObjectDetection, GroundingDinoProcessor
|
|
| 20 |
from diffusers import FluxPipeline
|
| 21 |
from transformers import pipeline, AutoTokenizer, AutoModelForSeq2SeqLM
|
| 22 |
import gc
|
| 23 |
-
import base64
|
| 24 |
|
| 25 |
-
|
| 26 |
-
# GPU ์ค์
|
| 27 |
-
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") # ๋ช
์์ ์ผ๋ก cuda:0 ์ง์
|
| 28 |
-
|
| 29 |
-
###--------------ZERO GPU ํ์/ ๋ฉ๋ชจ๋ฆฌ ๊ด๋ฆฌ ๊ณตํต --------------------###
|
| 30 |
def clear_memory():
|
| 31 |
"""๋ฉ๋ชจ๋ฆฌ ์ ๋ฆฌ ํจ์"""
|
| 32 |
gc.collect()
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
with torch.cuda.device(
|
| 36 |
torch.cuda.empty_cache()
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
print(f"Warning: Could not clear CUDA memory: {e}")
|
| 40 |
|
| 41 |
-
|
|
|
|
| 42 |
|
| 43 |
# GPU ์ค์ ์ try-except๋ก ๊ฐ์ธ๊ธฐ
|
| 44 |
if torch.cuda.is_available():
|
|
@@ -94,35 +88,14 @@ gd_model = GroundingDinoForObjectDetection.from_pretrained(gd_model_path, torch_
|
|
| 94 |
gd_model = gd_model.to(device=device)
|
| 95 |
assert isinstance(gd_model, GroundingDinoForObjectDetection)
|
| 96 |
|
| 97 |
-
# ํ์ดํ๋ผ์ธ ์ด๊ธฐํ
|
| 98 |
pipe = FluxPipeline.from_pretrained(
|
| 99 |
"black-forest-labs/FLUX.1-dev",
|
| 100 |
torch_dtype=torch.float16,
|
| 101 |
use_auth_token=HF_TOKEN
|
| 102 |
)
|
| 103 |
-
|
| 104 |
-
# ๋ฉ๋ชจ๋ฆฌ ์ต์ ํ ์ค์ - FluxPipeline์์ ์ง์ํ๋ ๋ฉ์๋๋ง ์ฌ์ฉ
|
| 105 |
pipe.enable_attention_slicing(slice_size="auto")
|
| 106 |
|
| 107 |
-
# xformers ์ต์ ํ (์ค์น๋์ด ์๋ ๊ฒฝ์ฐ์๋ง)
|
| 108 |
-
try:
|
| 109 |
-
import xformers
|
| 110 |
-
pipe.enable_xformers_memory_efficient_attention()
|
| 111 |
-
except ImportError:
|
| 112 |
-
print("xformers is not installed. Skipping memory efficient attention.")
|
| 113 |
-
|
| 114 |
-
# GPU ์ค์
|
| 115 |
-
if torch.cuda.is_available():
|
| 116 |
-
try:
|
| 117 |
-
pipe = pipe.to("cuda:0")
|
| 118 |
-
# CPU ์คํ๋ก๋ฉ์ด ์ง์๋๋ ๊ฒฝ์ฐ์๋ง ํ์ฑํ
|
| 119 |
-
if hasattr(pipe, 'enable_model_cpu_offload'):
|
| 120 |
-
pipe.enable_model_cpu_offload()
|
| 121 |
-
except Exception as e:
|
| 122 |
-
print(f"Warning: Could not move pipeline to CUDA: {str(e)}")
|
| 123 |
-
|
| 124 |
-
|
| 125 |
-
|
| 126 |
# LoRA ๊ฐ์ค์น ๋ก๋
|
| 127 |
pipe.load_lora_weights(
|
| 128 |
hf_hub_download(
|
|
@@ -194,66 +167,56 @@ def apply_mask(img: Image.Image, mask_img: Image.Image, defringe: bool = True) -
|
|
| 194 |
return result
|
| 195 |
|
| 196 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 197 |
def calculate_dimensions(aspect_ratio: str, base_size: int = 512) -> tuple[int, int]:
|
| 198 |
"""์ ํ๋ ๋น์จ์ ๋ฐ๋ผ ์ด๋ฏธ์ง ํฌ๊ธฐ ๊ณ์ฐ"""
|
| 199 |
-
# FLUX ํ์ดํ๋ผ์ธ์ด ์ง์ํ๋ ์์ ํ ํฌ๊ธฐ ์ฌ์ฉ
|
| 200 |
if aspect_ratio == "1:1":
|
| 201 |
-
|
| 202 |
elif aspect_ratio == "16:9":
|
| 203 |
-
|
| 204 |
elif aspect_ratio == "9:16":
|
| 205 |
-
|
| 206 |
elif aspect_ratio == "4:3":
|
| 207 |
-
|
| 208 |
-
|
| 209 |
-
width = height = 512
|
| 210 |
-
|
| 211 |
-
# 8์ ๋ฐฐ์๋ก ์กฐ์
|
| 212 |
-
width = (width // 8) * 8
|
| 213 |
-
height = (height // 8) * 8
|
| 214 |
-
|
| 215 |
-
return width, height
|
| 216 |
|
|
|
|
| 217 |
def generate_background(prompt: str, aspect_ratio: str) -> Image.Image:
|
| 218 |
try:
|
| 219 |
-
# ์์ ํ ํฌ๊ธฐ ๊ณ์ฐ
|
| 220 |
width, height = calculate_dimensions(aspect_ratio)
|
|
|
|
| 221 |
|
| 222 |
-
|
| 223 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 224 |
with timer("Background generation"):
|
| 225 |
try:
|
| 226 |
-
# ๋จผ์ 512x512๋ก ์์ฑ
|
| 227 |
with torch.inference_mode():
|
| 228 |
image = pipe(
|
| 229 |
prompt=prompt,
|
| 230 |
-
width=
|
| 231 |
-
height=
|
| 232 |
num_inference_steps=8,
|
| 233 |
-
guidance_scale=4.0
|
| 234 |
).images[0]
|
| 235 |
-
|
| 236 |
-
# ์ํ๋ ํฌ๊ธฐ๋ก ๋ฆฌ์ฌ์ด์ฆ
|
| 237 |
-
if width != 512 or height != 512:
|
| 238 |
-
image = image.resize((width, height), Image.LANCZOS)
|
| 239 |
-
return image
|
| 240 |
-
|
| 241 |
except Exception as e:
|
| 242 |
print(f"Pipeline error: {str(e)}")
|
| 243 |
-
# ์๋ฌ ๋ฐ์ ์ ํฐ์ ๋ฐฐ๊ฒฝ ๋ฐํ
|
| 244 |
return Image.new('RGB', (width, height), 'white')
|
| 245 |
|
|
|
|
| 246 |
except Exception as e:
|
| 247 |
print(f"Background generation error: {str(e)}")
|
| 248 |
return Image.new('RGB', (512, 512), 'white')
|
| 249 |
|
| 250 |
-
|
| 251 |
-
def adjust_size_to_multiple_of_8(width: int, height: int) -> tuple[int, int]:
|
| 252 |
-
"""์ด๋ฏธ์ง ํฌ๊ธฐ๋ฅผ 8์ ๋ฐฐ์๋ก ์กฐ์ """
|
| 253 |
-
new_width = max(8, ((width + 7) // 8) * 8) # ์ต์ 8ํฝ์
๋ณด์ฅ
|
| 254 |
-
new_height = max(8, ((height + 7) // 8) * 8) # ์ต์ 8ํฝ์
๋ณด์ฅ
|
| 255 |
-
return new_width, new_height
|
| 256 |
-
|
| 257 |
def create_position_grid():
|
| 258 |
return """
|
| 259 |
<div class="position-grid" style="display: grid; grid-template-columns: repeat(3, 1fr); gap: 10px; width: 150px; margin: auto;">
|
|
@@ -310,24 +273,26 @@ def combine_with_background(foreground: Image.Image, background: Image.Image,
|
|
| 310 |
result.paste(scaled_foreground, (x, y), scaled_foreground)
|
| 311 |
return result
|
| 312 |
|
| 313 |
-
@spaces.GPU(duration=
|
| 314 |
def _gpu_process(img: Image.Image, prompt: str | BoundingBox | None) -> tuple[Image.Image, BoundingBox | None, list[str]]:
|
|
|
|
| 315 |
try:
|
| 316 |
-
|
| 317 |
-
|
| 318 |
-
|
| 319 |
-
|
| 320 |
-
|
| 321 |
-
|
| 322 |
-
|
| 323 |
-
|
| 324 |
-
|
| 325 |
-
|
|
|
|
|
|
|
|
|
|
| 326 |
except Exception as e:
|
| 327 |
print(f"GPU process error: {str(e)}")
|
| 328 |
raise
|
| 329 |
-
finally:
|
| 330 |
-
clear_memory()
|
| 331 |
|
| 332 |
def _process(img: Image.Image, prompt: str | BoundingBox | None, bg_prompt: str | None = None, aspect_ratio: str = "1:1") -> tuple[tuple[Image.Image, Image.Image, Image.Image], gr.DownloadButton]:
|
| 333 |
try:
|
|
@@ -338,12 +303,16 @@ def _process(img: Image.Image, prompt: str | BoundingBox | None, bg_prompt: str
|
|
| 338 |
new_size = (int(img.width * ratio), int(img.height * ratio))
|
| 339 |
img = img.resize(new_size, Image.LANCZOS)
|
| 340 |
|
| 341 |
-
# CUDA ๋ฉ๋ชจ๋ฆฌ ๊ด๋ฆฌ
|
| 342 |
-
|
| 343 |
-
torch.cuda.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 344 |
|
| 345 |
-
|
| 346 |
-
with torch.amp.autocast('cuda', enabled=torch.cuda.is_available()):
|
| 347 |
mask, bbox, time_log = _gpu_process(img, prompt)
|
| 348 |
masked_alpha = apply_mask(img, mask, defringe=True)
|
| 349 |
|
|
@@ -376,42 +345,37 @@ def process_prompt(img: Image.Image, prompt: str, bg_prompt: str | None = None,
|
|
| 376 |
aspect_ratio: str = "1:1", position: str = "bottom-center",
|
| 377 |
scale_percent: float = 100) -> tuple[Image.Image, Image.Image]:
|
| 378 |
try:
|
| 379 |
-
if img is None or
|
| 380 |
raise gr.Error("Please provide both image and prompt")
|
| 381 |
|
| 382 |
-
print(f"Processing with position: {position}, scale: {scale_percent}")
|
| 383 |
|
| 384 |
-
|
| 385 |
-
|
| 386 |
-
|
| 387 |
-
|
| 388 |
-
|
|
|
|
| 389 |
|
| 390 |
-
|
| 391 |
-
translated_prompt = translate_to_english(prompt)
|
| 392 |
-
translated_bg_prompt = translate_to_english(bg_prompt) if bg_prompt else None
|
| 393 |
|
| 394 |
-
|
| 395 |
-
|
| 396 |
-
|
| 397 |
-
|
| 398 |
-
|
| 399 |
-
|
| 400 |
-
|
| 401 |
-
|
| 402 |
-
|
| 403 |
-
|
| 404 |
-
|
| 405 |
-
|
| 406 |
-
|
| 407 |
-
|
| 408 |
-
|
| 409 |
-
return results[1], results[2]
|
| 410 |
-
|
| 411 |
-
return results[1], results[2]
|
| 412 |
-
|
| 413 |
except Exception as e:
|
| 414 |
-
print(f"
|
| 415 |
raise gr.Error(str(e))
|
| 416 |
finally:
|
| 417 |
clear_memory()
|
|
@@ -518,61 +482,9 @@ button.primary:hover {
|
|
| 518 |
}
|
| 519 |
"""
|
| 520 |
|
|
|
|
|
|
|
| 521 |
|
| 522 |
-
def get_image_base64(image_path):
|
| 523 |
-
with open(image_path, "rb") as image_file:
|
| 524 |
-
return base64.b64encode(image_file.read()).decode()
|
| 525 |
-
|
| 526 |
-
# ์ด๋ฏธ์ง๋ฅผ Base64๋ก ๋ณํ
|
| 527 |
-
try:
|
| 528 |
-
example_img1 = get_image_base64("aa1.png")
|
| 529 |
-
example_img2 = get_image_base64("aa2.png")
|
| 530 |
-
example_img3 = get_image_base64("aa3.png")
|
| 531 |
-
except Exception as e:
|
| 532 |
-
print(f"Error loading example images: {e}")
|
| 533 |
-
example_img1 = example_img2 = example_img3 = ""
|
| 534 |
-
|
| 535 |
-
# HTML ํ
ํ๋ฆฟ ์์
|
| 536 |
-
example_html = f"""
|
| 537 |
-
<div style="margin-top: 50px; padding: 20px; background-color: #f8f9fa; border-radius: 10px;">
|
| 538 |
-
<h2 style="text-align: center; color: #2196F3; margin-bottom: 30px;">How It Works: Step by Step Guide</h2>
|
| 539 |
-
|
| 540 |
-
<div style="display: flex; justify-content: space-around; align-items: center; flex-wrap: wrap; gap: 20px;">
|
| 541 |
-
<div style="text-align: center; flex: 1; min-width: 250px; max-width: 300px;">
|
| 542 |
-
<img src="data:image/png;base64,{example_img1}"
|
| 543 |
-
style="width: 100%; border-radius: 8px; box-shadow: 0 2px 4px rgba(0,0,0,0.1);">
|
| 544 |
-
<h3 style="color: #333; margin: 15px 0;">Step 1: Original Image</h3>
|
| 545 |
-
<p style="color: #666;">Upload your original image containing the object you want to extract.</p>
|
| 546 |
-
</div>
|
| 547 |
-
|
| 548 |
-
<div style="text-align: center; flex: 1; min-width: 250px; max-width: 300px;">
|
| 549 |
-
<img src="data:image/png;base64,{example_img2}"
|
| 550 |
-
style="width: 100%; border-radius: 8px; box-shadow: 0 2px 4px rgba(0,0,0,0.1);">
|
| 551 |
-
<h3 style="color: #333; margin: 15px 0;">Step 2: Object Extraction</h3>
|
| 552 |
-
<p style="color: #666;">AI automatically detects and extracts the specified object.</p>
|
| 553 |
-
</div>
|
| 554 |
-
|
| 555 |
-
<div style="text-align: center; flex: 1; min-width: 250px; max-width: 300px;">
|
| 556 |
-
<img src="data:image/png;base64,{example_img3}"
|
| 557 |
-
style="width: 100%; border-radius: 8px; box-shadow: 0 2px 4px rgba(0,0,0,0.1);">
|
| 558 |
-
<h3 style="color: #333; margin: 15px 0;">Step 3: Final Result</h3>
|
| 559 |
-
<p style="color: #666;">The extracted object is placed on an AI-generated background.</p>
|
| 560 |
-
</div>
|
| 561 |
-
</div>
|
| 562 |
-
|
| 563 |
-
<div style="margin-top: 30px; text-align: center; padding: 20px; background-color: #e3f2fd; border-radius: 8px;">
|
| 564 |
-
<h4 style="color: #1976D2; margin-bottom: 10px;">Key Features:</h4>
|
| 565 |
-
<ul style="list-style: none; padding: 0;">
|
| 566 |
-
<li style="margin: 5px 0;">โจ Advanced AI-powered object detection and extraction</li>
|
| 567 |
-
<li style="margin: 5px 0;">๐จ Custom background generation with text prompts</li>
|
| 568 |
-
<li style="margin: 5px 0;">๐ Flexible object positioning and sizing options</li>
|
| 569 |
-
<li style="margin: 5px 0;">๐ Multiple aspect ratio support for various use cases</li>
|
| 570 |
-
</ul>
|
| 571 |
-
</div>
|
| 572 |
-
</div>
|
| 573 |
-
"""
|
| 574 |
-
|
| 575 |
-
|
| 576 |
# UI ๊ตฌ์ฑ
|
| 577 |
with gr.Blocks(theme=gr.themes.Soft(), css=css) as demo:
|
| 578 |
gr.HTML("""
|
|
@@ -582,9 +494,6 @@ with gr.Blocks(theme=gr.themes.Soft(), css=css) as demo:
|
|
| 582 |
</div>
|
| 583 |
""")
|
| 584 |
|
| 585 |
-
# ์์ ์น์
์ถ๊ฐ
|
| 586 |
-
gr.HTML(example_html)
|
| 587 |
-
|
| 588 |
with gr.Row():
|
| 589 |
with gr.Column(scale=1):
|
| 590 |
input_image = gr.Image(
|
|
@@ -632,7 +541,7 @@ with gr.Blocks(theme=gr.themes.Soft(), css=css) as demo:
|
|
| 632 |
scale_slider = gr.Slider(
|
| 633 |
minimum=10,
|
| 634 |
maximum=200,
|
| 635 |
-
value=
|
| 636 |
step=5,
|
| 637 |
label="Object Size (%)"
|
| 638 |
)
|
|
@@ -689,13 +598,12 @@ with gr.Blocks(theme=gr.themes.Soft(), css=css) as demo:
|
|
| 689 |
)
|
| 690 |
|
| 691 |
def update_controls(bg_prompt):
|
|
|
|
| 692 |
is_visible = bool(bg_prompt)
|
| 693 |
return [
|
| 694 |
-
gr.update(visible=is_visible
|
| 695 |
gr.update(visible=is_visible), # object_controls
|
| 696 |
]
|
| 697 |
-
|
| 698 |
-
|
| 699 |
|
| 700 |
bg_prompt.change(
|
| 701 |
fn=update_controls,
|
|
@@ -724,4 +632,5 @@ demo.launch(
|
|
| 724 |
server_name="0.0.0.0",
|
| 725 |
server_port=7860,
|
| 726 |
share=False,
|
| 727 |
-
max_threads=2
|
|
|
|
|
|
| 20 |
from diffusers import FluxPipeline
|
| 21 |
from transformers import pipeline, AutoTokenizer, AutoModelForSeq2SeqLM
|
| 22 |
import gc
|
|
|
|
| 23 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
def clear_memory():
|
| 25 |
"""๋ฉ๋ชจ๋ฆฌ ์ ๋ฆฌ ํจ์"""
|
| 26 |
gc.collect()
|
| 27 |
+
try:
|
| 28 |
+
if torch.cuda.is_available():
|
| 29 |
+
with torch.cuda.device(0): # ๋ช
์์ ์ผ๋ก device 0 ์ฌ์ฉ
|
| 30 |
torch.cuda.empty_cache()
|
| 31 |
+
except:
|
| 32 |
+
pass
|
|
|
|
| 33 |
|
| 34 |
+
# GPU ์ค์
|
| 35 |
+
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") # ๋ช
์์ ์ผ๋ก cuda:0 ์ง์
|
| 36 |
|
| 37 |
# GPU ์ค์ ์ try-except๋ก ๊ฐ์ธ๊ธฐ
|
| 38 |
if torch.cuda.is_available():
|
|
|
|
| 88 |
gd_model = gd_model.to(device=device)
|
| 89 |
assert isinstance(gd_model, GroundingDinoForObjectDetection)
|
| 90 |
|
| 91 |
+
# FLUX ํ์ดํ๋ผ์ธ ์ด๊ธฐํ
|
| 92 |
pipe = FluxPipeline.from_pretrained(
|
| 93 |
"black-forest-labs/FLUX.1-dev",
|
| 94 |
torch_dtype=torch.float16,
|
| 95 |
use_auth_token=HF_TOKEN
|
| 96 |
)
|
|
|
|
|
|
|
| 97 |
pipe.enable_attention_slicing(slice_size="auto")
|
| 98 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 99 |
# LoRA ๊ฐ์ค์น ๋ก๋
|
| 100 |
pipe.load_lora_weights(
|
| 101 |
hf_hub_download(
|
|
|
|
| 167 |
return result
|
| 168 |
|
| 169 |
|
| 170 |
+
def adjust_size_to_multiple_of_8(width: int, height: int) -> tuple[int, int]:
|
| 171 |
+
"""์ด๋ฏธ์ง ํฌ๊ธฐ๋ฅผ 8์ ๋ฐฐ์๋ก ์กฐ์ ํ๋ ํจ์"""
|
| 172 |
+
new_width = ((width + 7) // 8) * 8
|
| 173 |
+
new_height = ((height + 7) // 8) * 8
|
| 174 |
+
return new_width, new_height
|
| 175 |
+
|
| 176 |
def calculate_dimensions(aspect_ratio: str, base_size: int = 512) -> tuple[int, int]:
|
| 177 |
"""์ ํ๋ ๋น์จ์ ๋ฐ๋ผ ์ด๋ฏธ์ง ํฌ๊ธฐ ๊ณ์ฐ"""
|
|
|
|
| 178 |
if aspect_ratio == "1:1":
|
| 179 |
+
return base_size, base_size
|
| 180 |
elif aspect_ratio == "16:9":
|
| 181 |
+
return base_size * 16 // 9, base_size
|
| 182 |
elif aspect_ratio == "9:16":
|
| 183 |
+
return base_size, base_size * 16 // 9
|
| 184 |
elif aspect_ratio == "4:3":
|
| 185 |
+
return base_size * 4 // 3, base_size
|
| 186 |
+
return base_size, base_size
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 187 |
|
| 188 |
+
@spaces.GPU(duration=20) # 40์ด์์ 20์ด๋ก ๊ฐ์
|
| 189 |
def generate_background(prompt: str, aspect_ratio: str) -> Image.Image:
|
| 190 |
try:
|
|
|
|
| 191 |
width, height = calculate_dimensions(aspect_ratio)
|
| 192 |
+
width, height = adjust_size_to_multiple_of_8(width, height)
|
| 193 |
|
| 194 |
+
max_size = 768
|
| 195 |
+
if width > max_size or height > max_size:
|
| 196 |
+
ratio = max_size / max(width, height)
|
| 197 |
+
width = int(width * ratio)
|
| 198 |
+
height = int(height * ratio)
|
| 199 |
+
width, height = adjust_size_to_multiple_of_8(width, height)
|
| 200 |
+
|
| 201 |
with timer("Background generation"):
|
| 202 |
try:
|
|
|
|
| 203 |
with torch.inference_mode():
|
| 204 |
image = pipe(
|
| 205 |
prompt=prompt,
|
| 206 |
+
width=width,
|
| 207 |
+
height=height,
|
| 208 |
num_inference_steps=8,
|
| 209 |
+
guidance_scale=4.0
|
| 210 |
).images[0]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 211 |
except Exception as e:
|
| 212 |
print(f"Pipeline error: {str(e)}")
|
|
|
|
| 213 |
return Image.new('RGB', (width, height), 'white')
|
| 214 |
|
| 215 |
+
return image
|
| 216 |
except Exception as e:
|
| 217 |
print(f"Background generation error: {str(e)}")
|
| 218 |
return Image.new('RGB', (512, 512), 'white')
|
| 219 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 220 |
def create_position_grid():
|
| 221 |
return """
|
| 222 |
<div class="position-grid" style="display: grid; grid-template-columns: repeat(3, 1fr); gap: 10px; width: 150px; margin: auto;">
|
|
|
|
| 273 |
result.paste(scaled_foreground, (x, y), scaled_foreground)
|
| 274 |
return result
|
| 275 |
|
| 276 |
+
@spaces.GPU(duration=30) # 120์ด์์ 30์ด๋ก ๊ฐ์
|
| 277 |
def _gpu_process(img: Image.Image, prompt: str | BoundingBox | None) -> tuple[Image.Image, BoundingBox | None, list[str]]:
|
| 278 |
+
time_log: list[str] = []
|
| 279 |
try:
|
| 280 |
+
if isinstance(prompt, str):
|
| 281 |
+
t0 = time.time()
|
| 282 |
+
bbox = gd_detect(img, prompt)
|
| 283 |
+
time_log.append(f"detect: {time.time() - t0}")
|
| 284 |
+
if not bbox:
|
| 285 |
+
print(time_log[0])
|
| 286 |
+
raise gr.Error("No object detected")
|
| 287 |
+
else:
|
| 288 |
+
bbox = prompt
|
| 289 |
+
t0 = time.time()
|
| 290 |
+
mask = segmenter(img, bbox)
|
| 291 |
+
time_log.append(f"segment: {time.time() - t0}")
|
| 292 |
+
return mask, bbox, time_log
|
| 293 |
except Exception as e:
|
| 294 |
print(f"GPU process error: {str(e)}")
|
| 295 |
raise
|
|
|
|
|
|
|
| 296 |
|
| 297 |
def _process(img: Image.Image, prompt: str | BoundingBox | None, bg_prompt: str | None = None, aspect_ratio: str = "1:1") -> tuple[tuple[Image.Image, Image.Image, Image.Image], gr.DownloadButton]:
|
| 298 |
try:
|
|
|
|
| 303 |
new_size = (int(img.width * ratio), int(img.height * ratio))
|
| 304 |
img = img.resize(new_size, Image.LANCZOS)
|
| 305 |
|
| 306 |
+
# CUDA ๋ฉ๋ชจ๋ฆฌ ๊ด๋ฆฌ ์์
|
| 307 |
+
try:
|
| 308 |
+
if torch.cuda.is_available():
|
| 309 |
+
current_device = torch.cuda.current_device()
|
| 310 |
+
with torch.cuda.device(current_device):
|
| 311 |
+
torch.cuda.empty_cache()
|
| 312 |
+
except Exception as e:
|
| 313 |
+
print(f"CUDA memory management failed: {e}")
|
| 314 |
|
| 315 |
+
with torch.cuda.amp.autocast(enabled=torch.cuda.is_available()):
|
|
|
|
| 316 |
mask, bbox, time_log = _gpu_process(img, prompt)
|
| 317 |
masked_alpha = apply_mask(img, mask, defringe=True)
|
| 318 |
|
|
|
|
| 345 |
aspect_ratio: str = "1:1", position: str = "bottom-center",
|
| 346 |
scale_percent: float = 100) -> tuple[Image.Image, Image.Image]:
|
| 347 |
try:
|
| 348 |
+
if img is None or prompt.strip() == "":
|
| 349 |
raise gr.Error("Please provide both image and prompt")
|
| 350 |
|
| 351 |
+
print(f"Processing with position: {position}, scale: {scale_percent}")
|
| 352 |
|
| 353 |
+
try:
|
| 354 |
+
prompt = translate_to_english(prompt)
|
| 355 |
+
if bg_prompt:
|
| 356 |
+
bg_prompt = translate_to_english(bg_prompt)
|
| 357 |
+
except Exception as e:
|
| 358 |
+
print(f"Translation error (continuing with original text): {str(e)}")
|
| 359 |
|
| 360 |
+
results, _ = _process(img, prompt, bg_prompt, aspect_ratio)
|
|
|
|
|
|
|
| 361 |
|
| 362 |
+
if bg_prompt:
|
| 363 |
+
try:
|
| 364 |
+
combined = combine_with_background(
|
| 365 |
+
foreground=results[2],
|
| 366 |
+
background=results[1],
|
| 367 |
+
position=position,
|
| 368 |
+
scale_percent=scale_percent
|
| 369 |
+
)
|
| 370 |
+
print(f"Combined image created with position: {position}")
|
| 371 |
+
return combined, results[2]
|
| 372 |
+
except Exception as e:
|
| 373 |
+
print(f"Combination error: {str(e)}")
|
| 374 |
+
return results[1], results[2]
|
| 375 |
+
|
| 376 |
+
return results[1], results[2]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 377 |
except Exception as e:
|
| 378 |
+
print(f"Error in process_prompt: {str(e)}")
|
| 379 |
raise gr.Error(str(e))
|
| 380 |
finally:
|
| 381 |
clear_memory()
|
|
|
|
| 482 |
}
|
| 483 |
"""
|
| 484 |
|
| 485 |
+
# UI ๊ตฌ์ฑ
|
| 486 |
+
# UI ๊ตฌ์ฑ ๋ถ๋ถ์์ process_btn์ ์๋ก ์ด๋ํ๊ณ position_grid.click ๋ถ๋ถ ์ ๊ฑฐ
|
| 487 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 488 |
# UI ๊ตฌ์ฑ
|
| 489 |
with gr.Blocks(theme=gr.themes.Soft(), css=css) as demo:
|
| 490 |
gr.HTML("""
|
|
|
|
| 494 |
</div>
|
| 495 |
""")
|
| 496 |
|
|
|
|
|
|
|
|
|
|
| 497 |
with gr.Row():
|
| 498 |
with gr.Column(scale=1):
|
| 499 |
input_image = gr.Image(
|
|
|
|
| 541 |
scale_slider = gr.Slider(
|
| 542 |
minimum=10,
|
| 543 |
maximum=200,
|
| 544 |
+
value=50,
|
| 545 |
step=5,
|
| 546 |
label="Object Size (%)"
|
| 547 |
)
|
|
|
|
| 598 |
)
|
| 599 |
|
| 600 |
def update_controls(bg_prompt):
|
| 601 |
+
"""๋ฐฐ๊ฒฝ ํ๋กฌํํธ ์
๋ ฅ ์ฌ๋ถ์ ๋ฐ๋ผ ์ปจํธ๋กค ํ์ ์
๋ฐ์ดํธ"""
|
| 602 |
is_visible = bool(bg_prompt)
|
| 603 |
return [
|
| 604 |
+
gr.update(visible=is_visible), # aspect_ratio
|
| 605 |
gr.update(visible=is_visible), # object_controls
|
| 606 |
]
|
|
|
|
|
|
|
| 607 |
|
| 608 |
bg_prompt.change(
|
| 609 |
fn=update_controls,
|
|
|
|
| 632 |
server_name="0.0.0.0",
|
| 633 |
server_port=7860,
|
| 634 |
share=False,
|
| 635 |
+
max_threads=2 # ์ค๋ ๋ ์ ์ ํ
|
| 636 |
+
)
|