Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,41 +1,38 @@
|
|
|
|
1 |
from PIL import Image
|
2 |
-
from transformers import VisionEncoderDecoderModel, ViTFeatureExtractor, PreTrainedTokenizerFast
|
3 |
import gradio as gr
|
4 |
|
5 |
-
# Load the
|
6 |
-
|
7 |
-
|
8 |
-
tokenizer = PreTrainedTokenizerFast.from_pretrained("microsoft/git-base")
|
9 |
|
10 |
# Define the captioning function
|
11 |
-
def
|
12 |
-
#
|
13 |
-
pixel_values =
|
14 |
# Generate captions
|
15 |
-
|
16 |
-
|
17 |
-
return
|
18 |
|
19 |
# Define Gradio interface components
|
20 |
inputs = [
|
21 |
-
gr.inputs.Image(type='pil', label='
|
22 |
]
|
23 |
|
24 |
outputs = [
|
25 |
-
gr.outputs.Textbox(label='Caption')
|
26 |
]
|
27 |
|
28 |
# Define Gradio app properties
|
29 |
-
title = "
|
30 |
-
description = "Upload an image to see the caption generated"
|
31 |
-
example = ['messi.jpg'] # Replace with a valid path to an example image
|
32 |
|
33 |
# Create and launch the Gradio interface
|
34 |
gr.Interface(
|
35 |
-
fn=
|
36 |
inputs=inputs,
|
37 |
outputs=outputs,
|
38 |
title=title,
|
39 |
description=description,
|
40 |
-
examples=example,
|
41 |
).launch(debug=True)
|
|
|
1 |
+
from transformers import AutoProcessor, AutoModelForCausalLM
|
2 |
from PIL import Image
|
|
|
3 |
import gradio as gr
|
4 |
|
5 |
+
# Load the processor and model
|
6 |
+
processor = AutoProcessor.from_pretrained("microsoft/git-base-coco")
|
7 |
+
model = AutoModelForCausalLM.from_pretrained("microsoft/git-base-coco")
|
|
|
8 |
|
9 |
# Define the captioning function
|
10 |
+
def caption_image(image):
|
11 |
+
# Process the image
|
12 |
+
pixel_values = processor(images=image, return_tensors="pt").pixel_values
|
13 |
# Generate captions
|
14 |
+
generated_ids = model.generate(pixel_values=pixel_values, max_length=50)
|
15 |
+
generated_caption = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
16 |
+
return generated_caption
|
17 |
|
18 |
# Define Gradio interface components
|
19 |
inputs = [
|
20 |
+
gr.inputs.Image(type='pil', label='Upload Image')
|
21 |
]
|
22 |
|
23 |
outputs = [
|
24 |
+
gr.outputs.Textbox(label='Generated Caption')
|
25 |
]
|
26 |
|
27 |
# Define Gradio app properties
|
28 |
+
title = "Image Captioning Application"
|
29 |
+
description = "Upload an image to see the caption generated by the model"
|
|
|
30 |
|
31 |
# Create and launch the Gradio interface
|
32 |
gr.Interface(
|
33 |
+
fn=caption_image,
|
34 |
inputs=inputs,
|
35 |
outputs=outputs,
|
36 |
title=title,
|
37 |
description=description,
|
|
|
38 |
).launch(debug=True)
|