Spaces:
Running
Running
import streamlit as st | |
import numpy as np | |
from PIL import Image | |
from transformers import AutoTokenizer, ViTFeatureExtractor, VisionEncoderDecoderModel, GPT2Tokenizer, GPT2LMHeadModel | |
import torch | |
from transformers import BartTokenizer, BartForConditionalGeneration | |
# Load pre-trained BART model and tokenizer | |
tokenizer_2 = BartTokenizer.from_pretrained("facebook/bart-large-cnn") | |
model_2 = BartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn") | |
# Directory path to the saved model on Google Drive | |
model = VisionEncoderDecoderModel.from_pretrained("nlpconnect/vit-gpt2-image-captioning") | |
feature_extractor = ViTFeatureExtractor.from_pretrained("nlpconnect/vit-gpt2-image-captioning") | |
tokenizer = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning") | |
def generate_captions(image): | |
image = Image.open(image).convert("RGB") | |
generated_caption = tokenizer.decode(model.generate(feature_extractor(image, return_tensors="pt").pixel_values.to("cpu"))[0]) | |
sentence = generated_caption | |
text_to_remove = "<|endoftext|>" | |
generated_caption = sentence.replace(text_to_remove, "") | |
return generated_caption | |
def generate_paragraph(caption): | |
# Tokenize the caption | |
inputs = tokenizer_2([caption], max_length=1024, truncation=True, padding="longest", return_tensors="pt") | |
# Generate text | |
output = model_2.generate(inputs.input_ids, attention_mask=inputs.attention_mask, max_length=200, num_beams=4, length_penalty=2.0, early_stopping=True) | |
# Decode the generated output | |
generated_text = tokenizer_2.decode(output[0], skip_special_tokens=True) | |
return generated_text | |
# create the Streamlit app | |
def app(): | |
st.title('Image from your Side, Detailed description from my site') | |
st.write('Upload an image to see what we have in store.') | |
# create file uploader | |
uploaded_file = st.file_uploader("Got You Covered, Upload your wish!, magic on the Way! ", type=["jpg", "jpeg", "png"]) | |
# check if file has been uploaded | |
if uploaded_file is not None: | |
# load the image | |
image = Image.open(uploaded_file).convert("RGB") | |
# Image Captions | |
string = generate_captions(uploaded_file) | |
st.image(image, caption='The Uploaded File') | |
st.write("First is first captions for your Photo : ", string) | |
generated_paragraph = generate_paragraph(string) | |
st.write(generated_paragraph) | |
# run the app | |
if __name__ == '__main__': | |
app() |