Upload handler.py
#31
by
yanis9351
- opened
- handler.py +44 -0
handler.py
ADDED
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from typing import Dict, List, Any
|
2 |
+
from diffusers import StableDiffusionUpscalePipeline
|
3 |
+
import torch
|
4 |
+
from PIL import Image
|
5 |
+
import io
|
6 |
+
|
7 |
+
class EndpointHandler:
|
8 |
+
def __init__(self, path: str):
|
9 |
+
# Load the Stable Diffusion x4 upscaler model
|
10 |
+
self.pipeline = StableDiffusionUpscalePipeline.from_pretrained(
|
11 |
+
path,
|
12 |
+
torch_dtype=torch.float16
|
13 |
+
)
|
14 |
+
self.pipeline.to("cuda")
|
15 |
+
|
16 |
+
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
17 |
+
"""
|
18 |
+
data args:
|
19 |
+
inputs: str - The text prompt for the upscaling.
|
20 |
+
image: bytes - The low-resolution image as byte data.
|
21 |
+
|
22 |
+
Return:
|
23 |
+
A list of dictionaries with the upscaled image.
|
24 |
+
"""
|
25 |
+
# Extract inputs and image from the payload
|
26 |
+
prompt = data.get("inputs", "")
|
27 |
+
image_bytes = data.get("image", None)
|
28 |
+
|
29 |
+
if image_bytes is None:
|
30 |
+
return [{"error": "No image provided"}]
|
31 |
+
|
32 |
+
# Convert the byte data to an image
|
33 |
+
low_res_img = Image.open(io.BytesIO(image_bytes)).convert("RGB")
|
34 |
+
|
35 |
+
# Perform upscaling
|
36 |
+
upscaled_image = self.pipeline(prompt=prompt, image=low_res_img).images[0]
|
37 |
+
|
38 |
+
# Save the upscaled image to a byte stream
|
39 |
+
byte_io = io.BytesIO()
|
40 |
+
upscaled_image.save(byte_io, format="PNG")
|
41 |
+
byte_io.seek(0)
|
42 |
+
|
43 |
+
# Return the upscaled image as byte data
|
44 |
+
return [{"upscaled_image": byte_io.getvalue()}]
|