app file
Browse files
app.py
ADDED
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import (
|
2 |
+
TrOCRConfig,
|
3 |
+
TrOCRProcessor,
|
4 |
+
TrOCRForCausalLM,
|
5 |
+
ViTConfig,
|
6 |
+
ViTModel,
|
7 |
+
VisionEncoderDecoderModel,
|
8 |
+
)
|
9 |
+
import gradio as gr
|
10 |
+
|
11 |
+
processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-handwritten")
|
12 |
+
model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-handwritten")
|
13 |
+
def ocr(image):
|
14 |
+
pixel_values = processor(image, return_tensors="pt").pixel_values
|
15 |
+
generated_ids = model.generate(pixel_values)
|
16 |
+
generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
17 |
+
return generated_text
|
18 |
+
|
19 |
+
|
20 |
+
demo = gr.Interface(fn=ocr, inputs="image",outputs= ["text"])
|
21 |
+
demo.launch()
|