anamikau/Neural_Network

1import pandas as pd2import streamlit as st3import numpy as np4import tensorflow as tf5from PIL import Image6import pickle7 8 9 10st.header('Neural Networks Demo')11task = st.selectbox('Select Task', ["Select One",'Sentiment Classification', 'Tumor Detection'])12 13 14if task == "Tumor Detection":15        def cnn(img, model):16            img = Image.open(img)17            img = img.resize((128, 128))18            img = np.array(img)19            input_img = np.expand_dims(img, axis=0)20            res = model.predict(input_img)21            if res:22                return "Tumor Detected"23            else:24                return "No Tumor" 25            26        cnn_model = tf.keras.models.load_model("tumor_detection_model.h5")27        uploaded_file = st.file_uploader("Choose a file", type=["jpg", "jpeg", "png"])28        if uploaded_file is not None:29            st.image(uploaded_file, caption="Uploaded Image", use_column_width=True)30            if st.button("Submit"):31                result=cnn(uploaded_file, cnn_model)32                st.write(result)33 34        35elif task == "Sentiment Classification":36        types = ["Perceptron","BackPropagation", "RNN","DNN", "LSTM"]37        input_text2 = st.radio("Select", types, horizontal=True)38 39        if input_text2 == "Perceptron":40                with open("ppn_model.pkl",'rb') as file:41                    perceptron = pickle.load(file)42                with open("ppn_tokeniser.pkl",'rb') as file:43                    ppn_tokeniser = pickle.load(file)44 45                def ppn_make_predictions(inp, model):46                    encoded_inp = ppn_tokeniser.texts_to_sequences([inp])47                    padded_inp = tf.keras.preprocessing.sequence.pad_sequences(encoded_inp, maxlen=500)48                    res = model.predict(padded_inp)49                    if res:50                        return "Negative"51                    else:52                        return "Positive"       53                54                st.subheader('Movie Review Classification using Perceptron')55                inp = st.text_area('Enter message')56                if st.button('Check'):57                    pred = ppn_make_predictions([inp], perceptron)58                    st.write(pred)59 60        if input_text2 == "BackPropagation":61                with open("bp_model.pkl",'rb') as file:62                    backprop = pickle.load(file)63                with open("bp_tokeniser.pkl",'rb') as file:64                    bp_tokeniser = pickle.load(file)65 66                def bp_make_predictions(inp, model):67                    encoded_inp = bp_tokeniser.texts_to_sequences([inp])68                    padded_inp = tf.keras.preprocessing.sequence.pad_sequences(encoded_inp, maxlen=500)69                    res = model.predict(padded_inp)70                    if res:71                        return "Negative"72                    else:73                        return "Positive"     74                       75                st.subheader('Movie Review Classification using BackPropagation')76                inp = st.text_area('Enter message')77                if st.button('Check'):78                    pred = bp_make_predictions([inp], backprop)79                    st.write(pred)80        81 82        elif input_text2 == "RNN":83                rnn_model=tf.keras.models.load_model("spam_model.h5")84                with open("spam_tokeniser.pkl", 'rb') as model_file:85                    rnn_tokeniser=pickle.load(model_file)86 87                def rnn_make_predictions(inp, model):88                    encoded_inp = rnn_tokeniser.texts_to_sequences(inp)89                    padded_inp = tf.keras.preprocessing.sequence.pad_sequences(encoded_inp, maxlen=10, padding='post')90                    res = (model.predict(padded_inp) > 0.5).astype("int32")91                    if res:92                        return "Spam"93                    else:94                        return "Ham"95 96                st.subheader('Spam message Classification using RNN')97                input = st.text_area("Give message")98                if st.button('Check'):99                    pred = rnn_make_predictions([input], rnn_model)100                    st.write(pred)101 102 103 104        elif input_text2 == "DNN":105                        dnn_model=tf.keras.models.load_model("dnn_model.h5")106                        with open("dnn_tokeniser.pkl",'rb') as file:107                            dnn_tokeniser = pickle.load(file)108 109                        def dnn_make_predictions(inp, model):110                            inp = dnn_tokeniser.texts_to_sequences(inp)111                            inp = tf.keras.preprocessing.sequence.pad_sequences(inp, maxlen=500)112                            res = (model.predict(inp) > 0.5).astype("int32")113                            if res:114                                return "Negative"115                            else:116                                return "Positive"       117                        118                        st.subheader('Movie Review Classification using DNN')119                        inp = st.text_area('Enter message')120                        if st.button('Check'):121                            pred = dnn_make_predictions([inp], dnn_model)122                            st.write(pred)123 124                            125 126        elif input_text2 == "LSTM":127                lstm_model=tf.keras.models.load_model("lstm_model.h5") 128 129                with open("lstm_tokeniser.pkl",'rb') as file:130                    lstm_tokeniser = pickle.load(file)131 132                def lstm_make_predictions(inp, model):133                    inp = lstm_tokeniser.texts_to_sequences(inp)134                    inp = tf.keras.preprocessing.sequence.pad_sequences(inp, maxlen=500)135                    res = (model.predict(inp) > 0.5).astype("int32")136                    if res:137                        return "Negative"138                    else:139                        return "Positive"140                st.subheader('Movie Review Classification using LSTM')141                inp = st.text_area('Enter message')142                if st.button('Check'):143                    pred = lstm_make_predictions([inp], lstm_model)144                    st.write(pred)  145 146 147 148 149                150 151 152 153 154