semantic_search_loop.py

def semantic_search_loop(df_cached_embeddings, model, tokenizer):
    """
    This function takes a dataframe of cached embeddings, a model, and a tokenizer, and performs semantic search for
    images related to the user's search query.

    Args:
        df_cached_embeddings (pandas.DataFrame): A dataframe of cached image embeddings
        tokeniser: Tokeniser to use for pre-processing the text data
        model: model : A pre-trained CLIP model.

    Returns:
        None
    """
    # Prompt user to input a search query
    search = st.text_input("Describe some images you want to see..", 'A band doing a photo shoot outside')

    # Calculate the text embedding for the search query using the provided model and tokenizer
    search_embedding = calculate_text_features([search], tokenizer, model, normalise=False).detach().numpy()[0]

    # Rank the cached image embeddings based on their similarity to the search query embedding
    df_results = rank_vectors(search_embedding, df_cached_embeddings.image_clip_rep, df_cached_embeddings[['url']])

    # Display a grid of the top 10 images that are most similar to the search query
    display_image_grid(df_results.url.values[:10], 5, 300)
    return