app.py

import streamlit as st
#from transformers import CLIPModel, pipeline, CLIPImageProcessor 
from transformers import pipeline
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 = CLIPImageProcessor.from_pretrained(MODEL_name)

    return processor, model

def inference_clip(options, image, processor, model):
    
    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
col_l, col_r = st.columns(2)

#CLIP_large = load_clip(model_size='large')
model_name = "openai/clip-vit-large-patch14-336"
classifier = pipeline("zero-shot-image-classification", model = model_name)



#### Loading picture
with col_l:
    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
with col_l:
    default_options = 'There is a car, There is no car'
    options = st.text_input(label="Please enter the classes", value=default_options).split(',')
    
    #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, processor=clip_processor, model=clip_model)
        scores = classifier(image, 
                    candidate_labels = options)

        with col_r:
            #st.write(result)
            st.dataframe(scores)