app.py

from transformers import AutoTokenizer, ViTFeatureExtractor, VisionEncoderDecoderModel 
import os
import tensorflow as tf
os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0'

device='cpu'

model_id = "nttdataspain/vit-gpt2-stablediffusion2-lora"
model = VisionEncoderDecoderModel.from_pretrained(model_id)
tokenizer = AutoTokenizer.from_pretrained(model_id)
feature_extractor = ViTFeatureExtractor.from_pretrained(model_id)

def predict(image):
    img = image.convert('RGB')
    model.eval()
    pixel_values = feature_extractor(images=[img], return_tensors="pt").pixel_values
    with torch.no_grad():
        output_ids = model.generate(pixel_values, max_length=16, num_beams=4, return_dict_in_generate=True).sequences

    preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
    preds = [pred.strip() for pred in preds]
    return preds[0]
# def get_image(img):
#   image=pipe(img)
#   return image[0]['generated_text']
image=gr.Interface(predict,title='Image to text',inputs= gr.Image(label="Upload any Image", type = 'pil'),outputs='text').launch(share=True)