Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -7,6 +7,7 @@ import pandas as pd
|
|
| 7 |
from nltk.translate import bleu_score
|
| 8 |
from nltk.translate.bleu_score import SmoothingFunction
|
| 9 |
import torch
|
|
|
|
| 10 |
|
| 11 |
yolo_weights_path = "final_wts.pt"
|
| 12 |
|
|
@@ -183,3 +184,33 @@ logits_flattened = slogits.reshape(-1, slogits.shape[-1])
|
|
| 183 |
processor.batch_decode([logits_flattened.argmax(-1)], skip_special_tokens=True)
|
| 184 |
|
| 185 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
from nltk.translate import bleu_score
|
| 8 |
from nltk.translate.bleu_score import SmoothingFunction
|
| 9 |
import torch
|
| 10 |
+
import gradio as gr
|
| 11 |
|
| 12 |
yolo_weights_path = "final_wts.pt"
|
| 13 |
|
|
|
|
| 184 |
processor.batch_decode([logits_flattened.argmax(-1)], skip_special_tokens=True)
|
| 185 |
|
| 186 |
|
| 187 |
+
---------------------------------------
|
| 188 |
+
|
| 189 |
+
|
| 190 |
+
def gradio_inference(image_path):
|
| 191 |
+
"""
|
| 192 |
+
Function to handle inference and output the generated texts and final processed texts.
|
| 193 |
+
"""
|
| 194 |
+
df, bounding_path, tokens, logits, gen_texts = inference(image_path, debug=False, return_texts='final_v2')
|
| 195 |
+
|
| 196 |
+
# Convert the DataFrame for final texts to a readable format
|
| 197 |
+
final_texts = df.to_string(index=False)
|
| 198 |
+
|
| 199 |
+
# Convert the list of generated texts into a readable string
|
| 200 |
+
gen_texts_output = '\n'.join(gen_texts)
|
| 201 |
+
|
| 202 |
+
return gen_texts_output, final_texts
|
| 203 |
+
|
| 204 |
+
image_input = gr.inputs.Image(type="filepath", label="Upload Image")
|
| 205 |
+
generated_output = gr.outputs.Textbox(label="Generated Texts")
|
| 206 |
+
final_output = gr.outputs.Textbox(label="Final Processed Texts")
|
| 207 |
+
|
| 208 |
+
interface = gr.Interface(
|
| 209 |
+
fn=gradio_inference,
|
| 210 |
+
inputs=image_input,
|
| 211 |
+
outputs=[generated_output, final_output],
|
| 212 |
+
title="OCR using LLMs",
|
| 213 |
+
description="Upload an image and get generated and final processed texts",
|
| 214 |
+
)
|
| 215 |
+
|
| 216 |
+
interface.launch()
|