ripple-search / app.py
tensorkelechi's picture
Create app.py
44883c6 verified
raw
history blame
1.91 kB
import ripple
import streamlit as stl
from tqdm.auto import tqdm
# streamlit app
stl.set_page_config(
page_title="Ripple",
)
stl.title("ripple search")
stl.write(
"An app that uses text input to search for described images, using embeddings of selected image datasets. Uses contrastive learning models(CLIP) and the sentence transformers library"
)
stl.link_button(
label="link to github and full library code",
url="https://github.com/kelechi-c/ripple_net",
)
dataset = stl.selectbox(
"choose huggingface dataset(bgger datasets take more time to embed..)",
options=[
"huggan/wikiart(1k)",
"huggan/wikiart(11k)",
"zh-plus/tiny-imagenet(110k)",
"lambdalabs/naruto-blip-captions(1k)",
"detection-datasets/fashionpedia(45k)",
],
)
# initalized global variables
embedded_data = None
embedder = None
text_search = None
ret_images = []
scores = []
if dataset and stl.button("embed image dataset"):
with stl.spinner("Initializing and creating image embeddings from dataset"):
embedder = ripple.ImageEmbedder(
dataset, retrieval_type="text-image", dataset_type="huggingface"
)
embedded_data = embedder.create_embeddings(device="cpu")
stl.success("Sucessfully embedded and dcreated image index")
if embedded_data is not None:
text_search = ripple.TextSearch(embedded_data, embedder.embed_model)
stl.success("Initialized text search class")
search_term = stl.text_input("Text description/search for image")
if search_term:
with stl.spinner("retrieving images with description.."):
scores, ret_images = text_search.get_similar_images(
search_term, k_images=4)
stl.success(f"sucessfully retrieved {len(ret_images)}")
for count, score, image in tqdm(zip(range(len(ret_images)), scores, ret_images)):
stl.image(image["image"][count])
stl.write(score)