app.py

import streamlit as st
import os
import json

def batch(iterable, n=1):
    l = len(iterable)
    for ndx in range(0, l, n):
        yield iterable[ndx:min(ndx + n, l)]

data_path = "./data/TeTIm-Eval-Mini"
category_map = {
    "Digital art": "sampled_art_digital",
    "Digital sketches": "sampled_art_sketch",
    "Traditional art": "sampled_art_traditional",
    "Baroque paintings": "sampled_painting_baroque",
    "High renaissance paintings": "sampled_painting_high-renaissance",
    "Neoclassicism paintings": "sampled_painting_neoclassicism",
    "Animal photography": "sampled_photo_animal",
    "Food photography": "sampled_photo_food",
    "Landscape photography": "sampled_photo_landscape",
    "Portrait photography": "sampled_photo_person"
}
models_map = {
    "real": "Real picture",
    "DALLE2": "DALL-E 2",
    "Stable Diffusion": "Stable Diffusion",
    "Latent Diffusion": "Latent Diffusion",
    "craiyon": "Craiyon (ex dalle-mini)",
    "GLIDE": "GLIDE"
}

if "current" not in st.session_state:
    st.session_state["current"] = 1

st.header("TeTIm-Eval-mini showcase")
category = st.selectbox("Category", category_map.keys())

path = os.path.join(data_path, "real", category_map[category], "images")
real_images = sorted(os.listdir(path), key=lambda e: int(e.split(".")[0]))

coll, _, colc, _, colr = st.columns(5)

prev = coll.button("« Previous")
state = colc.empty()
next = colr.button("Next »")

if prev:
    st.session_state["current"] = max(st.session_state["current"] - 1, 1)

if next:
    st.session_state["current"] = min(st.session_state["current"] + 1, len(real_images))

state.write(f"{st.session_state['current']}/{len(real_images)}")

path = os.path.join(data_path, "real", category_map[category], "annotations")
annotations = sorted(os.listdir(path), key=lambda e: int(e.split(".")[0]))

with open(os.path.join(path, annotations[st.session_state["current"]-1])) as annotation_f:
    annotation = json.load(annotation_f)

st.subheader(annotation["caption"])

for (modell, modell_name), (modelr, modelr_name) in batch(list(models_map.items()), n=2):
    coll_title, colr_title = st.columns(2)
    coll_title.subheader(modell_name)
    colr_title.subheader(modelr_name)

    pathl = os.path.join(data_path, modell, category_map[category])
    if modell == "real": pathl = os.path.join(pathl, "images")
    imagesl = sorted(os.listdir(pathl), key=lambda e: int(e.split(".")[0]))

    pathr = os.path.join(data_path, modelr, category_map[category])
    if modelr == "real": pathr = os.path.join(pathr, "images")
    imagesr = sorted(os.listdir(pathr), key=lambda e: int(e.split(".")[0]))

    coll_image, colr_image = st.columns(2)
    coll_image.image(os.path.join(pathl, imagesl[st.session_state["current"]-1]))
    colr_image.image(os.path.join(pathr, imagesr[st.session_state["current"]-1]))