Update app.py
Browse files
app.py
CHANGED
|
@@ -1,22 +1,43 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
from PIL import Image
|
| 3 |
-
import pytesseract
|
| 4 |
import torch
|
| 5 |
from transformers import CLIPProcessor, CLIPModel
|
| 6 |
import openai
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
# Load CLIP model and processor from Hugging Face
|
| 9 |
clip_model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
|
| 10 |
clip_processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
|
| 11 |
|
| 12 |
-
#
|
| 13 |
-
import os
|
| 14 |
openai.api_key = os.getenv("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")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
-
# Main function
|
| 17 |
def generate_alt_text(image):
|
| 18 |
# Step 1: OCR
|
| 19 |
-
ocr_text =
|
| 20 |
|
| 21 |
# Step 2: Visual Feature Matching
|
| 22 |
candidate_descriptions = [
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
from PIL import Image
|
|
|
|
| 3 |
import torch
|
| 4 |
from transformers import CLIPProcessor, CLIPModel
|
| 5 |
import openai
|
| 6 |
+
import requests
|
| 7 |
+
import base64
|
| 8 |
+
import io
|
| 9 |
+
import os
|
| 10 |
|
| 11 |
# Load CLIP model and processor from Hugging Face
|
| 12 |
clip_model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
|
| 13 |
clip_processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
|
| 14 |
|
| 15 |
+
# Load API keys from environment variables
|
|
|
|
| 16 |
openai.api_key = os.getenv("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")
|
| 17 |
+
ocr_api_key = os.getenv("K86540556188957")
|
| 18 |
+
|
| 19 |
+
# Function to perform OCR using OCR.Space
|
| 20 |
+
def extract_text_with_ocr_space(image):
|
| 21 |
+
buffered = io.BytesIO()
|
| 22 |
+
image.save(buffered, format="JPEG")
|
| 23 |
+
img_str = base64.b64encode(buffered.getvalue()).decode()
|
| 24 |
+
|
| 25 |
+
response = requests.post(
|
| 26 |
+
"https://api.ocr.space/parse/image",
|
| 27 |
+
data={
|
| 28 |
+
"base64Image": f"data:image/jpeg;base64,{img_str}",
|
| 29 |
+
"language": "eng",
|
| 30 |
+
"isOverlayRequired": False,
|
| 31 |
+
"apikey": ocr_api_key
|
| 32 |
+
}
|
| 33 |
+
)
|
| 34 |
+
result = response.json()
|
| 35 |
+
return result['ParsedResults'][0]['ParsedText'] if "ParsedResults" in result else ""
|
| 36 |
|
| 37 |
+
# Main function to generate alt text
|
| 38 |
def generate_alt_text(image):
|
| 39 |
# Step 1: OCR
|
| 40 |
+
ocr_text = extract_text_with_ocr_space(image)
|
| 41 |
|
| 42 |
# Step 2: Visual Feature Matching
|
| 43 |
candidate_descriptions = [
|