Spaces:
Sleeping
Sleeping
File size: 1,754 Bytes
b1779fd 131383f b1779fd 21bab57 b1779fd 3711151 b1779fd 3711151 b1779fd 3ddadce 21bab57 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 |
import torch
from PIL import Image
from transformers import AutoProcessor, AutoModelForImageClassification
import gradio as gr
import pytesseract
def classify_meme(image: Image.Image):
# OCR: extract text from image
extracted_text = pytesseract.image_to_string(image)
# Process image with SigLIP2 model
inputs = processor(images=image, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model(**inputs)
probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
predictions = {labels[i]: float(probs[0][i]) for i in range(len(labels))}
return {
"Predictions": predictions,
"Extracted Text": extracted_text.strip()
}
# Load model and processor from Hugging Face
model = AutoModelForImageClassification.from_pretrained("google/siglip2-base-patch16-naflex")
processor = AutoProcessor.from_pretrained("google/siglip2-base-patch16-naflex")
labels = model.config.id2label
def classify_meme(image: Image.Image):
inputs = processor(images=image, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model(**inputs)
probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
predictions = {labels[i]: float(probs[0][i]) for i in range(len(labels))}
return predictions
# Gradio interface
demo = gr.Interface(
fn=classify_meme,
inputs=gr.Image(type="pil"),
outputs=[
gr.Label(num_top_classes=2, label="Predictions"),
gr.Textbox(label="Extracted Text")
],
title="Meme Classifier with OCR",
description="Upload a meme to classify its sentiment and extract text using OCR."
)
if __name__ == "__main__":
demo.launch(share = True)
|