app.py

import streamlit as st
#from transformers import pipeline
import matplotlib.pyplot as plt
import numpy as np
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.python.keras.layers import Dense, Flatten
from tensorflow.keras.models import Sequential
from tensorflow.keras.optimizers import Adam
from sklearn.neighbors import NearestNeighbors
#import pathlib
#import re
#import os
import PIL
import random

#pipe = pipeline("sentiment-analysis")
n_classes = 5749

button = st.button("Choose Random Image from training data")

feature_data = np.load('features.npy')
x_data = np.load('pictures.npy')
y_data = np.load('name_data.npy')

knn = NearestNeighbors(n_neighbors=11)
knn.fit(feature_data)

model = Sequential()

pretrained_model= tf.keras.applications.ResNet50(include_top=False,
  input_shape=(250,250,3),
  pooling='avg',classes=n_classes,
  weights='imagenet')
for layer in pretrained_model.layers:
  layer.trainable=False

model.add(pretrained_model)
model.build([None, 250, 250, 3])
uploaded_files = st.file_uploader('Upload your files')

if button:

    rand_idx = random.randint(0, 13232)
    res = knn.kneighbors([feature_data[rand_idx]], return_distance=False)[0]
    first = True 
    for idx, i in enumerate(res):
      if first:
        first = False 
        st.write("original image")
        st.image(x_data[i])
      else:
        st.write(f"{idx}-th similar image")
        st.image(x_data[i])

elif uploaded_files:
    image = PIL.Image.open(uploaded_files)
    img_array = np.array(image)
    img_array = np.resize(img_array, (250, 250, 3))
    img_array = img_array.reshape((1, 250, 250, 3))
    im_features = model.predict(img_array)
    #st.write(im_features)
    res = knn.kneighbors(im_features, return_distance=False)[0]
    first = True 
    for idx, i in enumerate(res):
      if first:
        first = False
        st.write("original image")
        st.image(image)
      else:
        st.write(f"{idx}-th similar image")
        st.image(x_data[i])