kenken999 commited on
Commit
1742fa2
·
1 Parent(s): 4d0d1d9
Files changed (1) hide show
  1. app.py +11 -66
app.py CHANGED
@@ -1,68 +1,13 @@
1
  import gradio as gr
2
- import subprocess
3
- import torch
4
- from PIL import Image
5
- from transformers import AutoProcessor, AutoModelForCausalLM
6
 
7
- # import os
8
- # import random
9
- # from gradio_client import Client
10
-
11
-
12
- subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
13
-
14
- # Initialize Florence model
15
- device = "cuda" if torch.cuda.is_available() else "cpu"
16
- florence_model = AutoModelForCausalLM.from_pretrained('microsoft/Florence-2-base', trust_remote_code=True).to(device).eval()
17
- florence_processor = AutoProcessor.from_pretrained('microsoft/Florence-2-base', trust_remote_code=True)
18
-
19
- # api_key = os.getenv("HF_READ_TOKEN")
20
-
21
- def generate_caption(image):
22
- if not isinstance(image, Image.Image):
23
- image = Image.fromarray(image)
24
-
25
- inputs = florence_processor(text="<MORE_DETAILED_CAPTION>", images=image, return_tensors="pt").to(device)
26
- generated_ids = florence_model.generate(
27
- input_ids=inputs["input_ids"],
28
- pixel_values=inputs["pixel_values"],
29
- max_new_tokens=1024,
30
- early_stopping=False,
31
- do_sample=False,
32
- num_beams=3,
33
- )
34
- generated_text = florence_processor.batch_decode(generated_ids, skip_special_tokens=False)[0]
35
- parsed_answer = florence_processor.post_process_generation(
36
- generated_text,
37
- task="<MORE_DETAILED_CAPTION>",
38
- image_size=(image.width, image.height)
39
- )
40
- prompt = parsed_answer["<MORE_DETAILED_CAPTION>"]
41
- print("\nGeneration completed!:"+ prompt)
42
- return prompt
43
- # yield prompt, None
44
- # image_path = generate_image(prompt,random.randint(0, 4294967296))
45
- # yield prompt, image_path
46
-
47
- # def generate_image(prompt, seed=42, width=1024, height=1024):
48
- # try:
49
- # result = Client("KingNish/Realtime-FLUX", hf_token=api_key).predict(
50
- # prompt=prompt,
51
- # seed=seed,
52
- # width=width,
53
- # height=height,
54
- # api_name="/generate_image"
55
- # )
56
- # # Extract the image path from the result tuple
57
- # image_path = result[0]
58
- # return image_path
59
- # except Exception as e:
60
- # raise Exception(f"Error generating image: {str(e)}")
61
-
62
- io = gr.Interface(generate_caption,
63
- inputs=[gr.Image(label="Input Image")],
64
- outputs = [gr.Textbox(label="Output Prompt", lines=2, show_copy_button = True),
65
- # gr.Image(label="Output Image")
66
- ]
67
- )
68
- io.launch(api=True)
 
1
  import gradio as gr
 
 
 
 
2
 
3
+ def describe_image(image):
4
+ return "This is a simple image processing API."
5
+
6
+ iface = gr.Interface(
7
+ fn=describe_image,
8
+ inputs=gr.Image(type="filepath"),
9
+ outputs="text",
10
+ title="Simple Image API",
11
+ description="Upload an image and get a simple description."
12
+ )
13
+ iface.launch(api=True)