app.py

import streamlit as st
from transformers import CLIPModel, CLIPProcessor
import torch
from PIL import Image

#################################
#### FUNCTIONS

def load_clip(model_size='large'):
    if model_size == 'base':
        MODEL_name = 'openai/clip-vit-base-patch32'
    elif model_size == 'large':
        MODEL_name = 'openai/clip-vit-large-patch14'
    
    model = CLIPModel.from_pretrained(MODEL_name)
    processor = CLIPProcessor.from_pretrained(MODEL_name)

    return processor, model

def inference_clip(options, image):
    
    inputs = processor(text= options, images=image, return_tensors="pt", padding=True)
    with torch.no_grad():
        outputs = model(**inputs)

        #logits_per_text = outputs.logits_per_text
        logits_per_image = outputs.logits_per_image  # this is the image-text similarity score
        probs = logits_per_image.softmax(dim=1)  # we can take the softmax to get the label probabilities

    max_prob_idx = torch.argmax(probs)
    max_prob_option = options[max_prob_idx]
    max_prob = probs[max_prob_idx].item()
    return max_prob_option 

#################################
#### LAYOUT

CLIP_large = load_clip(model_size='large')

picture_file = st.file_uploader("Picture :", type=["jpg", "jpeg", "png"])
if picture_file is not None:
    image = Image.open(picture_file)
    st.image(image, caption='Please upload an image of the damage', use_column_width=True)

#image
options = st.text_input(label="Please enter the classes", value="")
options = list(options)

# button to launch compute
if st.button("Compute"):
    clip_processor, clip_model = load_clip(model_size='large')
    result = inference_clip(options = options, image = image)
    st.write(result)