Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -2,9 +2,7 @@ from fastapi import FastAPI
|
|
2 |
from fastapi.responses import RedirectResponse
|
3 |
import gradio as gr
|
4 |
from PIL import Image
|
5 |
-
import io
|
6 |
import numpy as np
|
7 |
-
import easyocr
|
8 |
from transformers import pipeline
|
9 |
from gtts import gTTS
|
10 |
import tempfile
|
@@ -12,8 +10,8 @@ import os
|
|
12 |
|
13 |
app = FastAPI()
|
14 |
|
15 |
-
# OCR Reader
|
16 |
-
ocr_reader =
|
17 |
|
18 |
# Captioning and VQA Pipelines
|
19 |
caption_model = pipeline("image-to-text", model="nlpconnect/vit-gpt2-image-captioning")
|
@@ -24,6 +22,12 @@ def process_image_question(image: Image.Image, question: str):
|
|
24 |
return "No image uploaded.", None
|
25 |
|
26 |
try:
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
# Convert PIL image to numpy array
|
28 |
np_image = np.array(image)
|
29 |
|
|
|
2 |
from fastapi.responses import RedirectResponse
|
3 |
import gradio as gr
|
4 |
from PIL import Image
|
|
|
5 |
import numpy as np
|
|
|
6 |
from transformers import pipeline
|
7 |
from gtts import gTTS
|
8 |
import tempfile
|
|
|
10 |
|
11 |
app = FastAPI()
|
12 |
|
13 |
+
# OCR Reader (lazy import inside function to avoid ImportError on Spaces)
|
14 |
+
ocr_reader = None
|
15 |
|
16 |
# Captioning and VQA Pipelines
|
17 |
caption_model = pipeline("image-to-text", model="nlpconnect/vit-gpt2-image-captioning")
|
|
|
22 |
return "No image uploaded.", None
|
23 |
|
24 |
try:
|
25 |
+
# Import EasyOCR only when needed
|
26 |
+
global ocr_reader
|
27 |
+
if ocr_reader is None:
|
28 |
+
import easyocr
|
29 |
+
ocr_reader = easyocr.Reader(['en'], gpu=False)
|
30 |
+
|
31 |
# Convert PIL image to numpy array
|
32 |
np_image = np.array(image)
|
33 |
|