multimodalart HF Staff commited on
Commit
0974218
·
1 Parent(s): c85c7f3

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +30 -0
app.py ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import open_clip
2
+ from PIL import Image
3
+
4
+ import gradio as gr
5
+
6
+ model, _, transform = open_clip.create_model_and_transform(
7
+ model_name="coca_ViT-L-14",
8
+ pretrained="laion2B-s13B-b90k"
9
+ )
10
+
11
+ # load an image
12
+ def generate_caption(image):
13
+ im = image.convert("RGB")
14
+ # transform the image and add a batch size dimension
15
+ im = transform(im).unsqueeze(0)
16
+
17
+ generated = model.generate(im)
18
+ generated = generated.detach()
19
+
20
+ return(open_clip.decode(generated[0]))
21
+
22
+ with gr.Blocks() as demo:
23
+ gr.Markdown("## Captioning with OpenCLIP CoCa")
24
+ with gr.Row():
25
+ inp = gr.Image(label="Image to Caption", type="pil")
26
+ out = gr.Textbox(label="Caption")
27
+ btn = gr.Button("Generate caption")
28
+ btn.click(fn=generate_caption,inputs=inp, outputs=out)
29
+
30
+ demo.launch()