Spaces:
Runtime error
Runtime error
Amit Gazal
commited on
Commit
·
6a046fb
1
Parent(s):
7bce5c7
wip
Browse files- app.py +65 -0
- requirements.txt +7 -0
app.py
ADDED
|
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from PIL import Image
|
| 3 |
+
import matplotlib.pyplot as plt
|
| 4 |
+
import torch
|
| 5 |
+
from torchvision import transforms
|
| 6 |
+
from transformers import AutoModelForImageSegmentation
|
| 7 |
+
|
| 8 |
+
model = AutoModelForImageSegmentation.from_pretrained('briaai/RMBG-2.0', trust_remote_code=True)
|
| 9 |
+
torch.set_float32_matmul_precision(['high', 'highest'][0])
|
| 10 |
+
model.eval()
|
| 11 |
+
|
| 12 |
+
def remove_background(input_image, holiday, message):
|
| 13 |
+
image_size = (1024, 1024)
|
| 14 |
+
# Transform the input image
|
| 15 |
+
transform_image = transforms.Compose([
|
| 16 |
+
transforms.Resize(image_size),
|
| 17 |
+
transforms.ToTensor(),
|
| 18 |
+
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
|
| 19 |
+
])
|
| 20 |
+
|
| 21 |
+
# Process the image
|
| 22 |
+
input_tensor = transform_image(input_image).unsqueeze(0)
|
| 23 |
+
if torch.cuda.is_available():
|
| 24 |
+
input_tensor = input_tensor.to('cuda')
|
| 25 |
+
|
| 26 |
+
# Generate prediction
|
| 27 |
+
with torch.no_grad():
|
| 28 |
+
preds = model(input_tensor)[-1].sigmoid().cpu()
|
| 29 |
+
pred = preds[0].squeeze()
|
| 30 |
+
pred_pil = transforms.ToPILImage()(pred)
|
| 31 |
+
mask = pred_pil.resize(input_image.size)
|
| 32 |
+
|
| 33 |
+
# Create image without background
|
| 34 |
+
result_image = input_image.copy()
|
| 35 |
+
result_image.putalpha(mask)
|
| 36 |
+
|
| 37 |
+
# Create image with only background
|
| 38 |
+
only_background_image = input_image.copy()
|
| 39 |
+
inverted_mask = Image.eval(mask, lambda x: 255 - x) # Invert the mask
|
| 40 |
+
only_background_image.putalpha(inverted_mask)
|
| 41 |
+
|
| 42 |
+
first_output_image = result_image
|
| 43 |
+
second_output_image = only_background_image
|
| 44 |
+
third_output_image = result_image
|
| 45 |
+
|
| 46 |
+
return first_output_image, second_output_image, third_output_image
|
| 47 |
+
|
| 48 |
+
# Replace the demo interface
|
| 49 |
+
demo = gr.Interface(
|
| 50 |
+
fn=remove_background,
|
| 51 |
+
inputs=[
|
| 52 |
+
gr.Image(type="pil"),
|
| 53 |
+
gr.Text(label="Holiday (e.g. Christmas, New Year's, etc.)"),
|
| 54 |
+
gr.Text(label="Optional Message", placeholder="Enter your holiday message here...")
|
| 55 |
+
],
|
| 56 |
+
outputs=[
|
| 57 |
+
gr.Image(type="pil", label="First Output"),
|
| 58 |
+
gr.Image(type="pil", label="Second Output"),
|
| 59 |
+
gr.Image(type="pil", label="Third Output")
|
| 60 |
+
],
|
| 61 |
+
title="Background Removal Tool",
|
| 62 |
+
description="Upload an image to remove its background"
|
| 63 |
+
)
|
| 64 |
+
|
| 65 |
+
demo.launch()
|
requirements.txt
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
gradio
|
| 2 |
+
torch
|
| 3 |
+
torchvision
|
| 4 |
+
pillow
|
| 5 |
+
kornia
|
| 6 |
+
transformers
|
| 7 |
+
timm
|