import streamlit as st import os import numpy as np import pandas as pd import re import torch import torch.nn as nn from torch.utils.data import Dataset from transformers import AutoTokenizer, AutoModelForSequenceClassification import matplotlib.pyplot as plt import warnings warnings.filterwarnings("ignore") # Set page config st.set_page_config( page_title="Deteksi Alergen dalam Resep", page_icon="🍲", layout="wide" ) # Set device device = torch.device("cuda" if torch.cuda.is_available() else "cpu") # Clean text function def clean_text(text): # Convert dashes to spaces for better tokenization text = text.replace('--', ' ') # Basic cleaning text = re.sub(r"http\S+", "", text) text = re.sub('\n', ' ', text) text = re.sub("[^a-zA-Z0-9\s]", " ", text) text = re.sub(" {2,}", " ", text) text = text.strip() text = text.lower() return text # Define model for multilabel classification class MultilabelBertClassifier(nn.Module): def __init__(self, model_name, num_labels): super(MultilabelBertClassifier, self).__init__() self.bert = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=num_labels) # Replace the classification head with our own for multilabel self.bert.classifier = nn.Linear(self.bert.config.hidden_size, num_labels) def forward(self, input_ids, attention_mask): outputs = self.bert(input_ids=input_ids, attention_mask=attention_mask) return outputs.logits # Function to predict allergens in new recipes @st.cache_resource def load_model(): # Target columns target_columns = ['susu', 'kacang', 'telur', 'makanan_laut', 'gandum'] # Initialize tokenizer tokenizer = AutoTokenizer.from_pretrained('indobenchmark/indobert-base-p2') # Initialize model model = MultilabelBertClassifier('indobenchmark/indobert-base-p1', len(target_columns)) # Load model weights if available model_path = "model/alergen_model.pt" try: # Try to load the model checkpoint = torch.load(model_path, map_location=device) model.load_state_dict(checkpoint['model_state_dict']) st.success("Model berhasil dimuat!") except Exception as e: st.error(f"Error loading model: {str(e)}") st.warning("Model belum tersedia. Silakan latih model terlebih dahulu atau upload file model.") model.to(device) model.eval() return model, tokenizer, target_columns def predict_allergens(ingredients_text, model, tokenizer, target_columns, max_length=128): # Clean the text cleaned_text = clean_text(ingredients_text) # Tokenize encoding = tokenizer.encode_plus( cleaned_text, add_special_tokens=True, max_length=max_length, truncation=True, return_tensors='pt', padding='max_length' ) input_ids = encoding['input_ids'].to(device) attention_mask = encoding['attention_mask'].to(device) with torch.no_grad(): outputs = model(input_ids=input_ids, attention_mask=attention_mask) predictions = torch.sigmoid(outputs) predictions_prob = predictions.cpu().numpy()[0] predictions_binary = (predictions > 0.5).float().cpu().numpy()[0] result = {} for i, target in enumerate(target_columns): result[target] = { 'present': bool(predictions_binary[i]), 'probability': float(predictions_prob[i]) } return result # Main application def main(): st.title("Deteksi Alergen dalam Resep") st.markdown(""" Aplikasi ini menggunakan model IndoBERT untuk mendeteksi kemungkinan alergen dalam resep berdasarkan daftar bahan. Alergen yang diidentifikasi meliputi: - Susu - Kacang - Telur - Makanan Laut - Gandum """) # Sidebar for model upload st.sidebar.header("Upload Model") uploaded_model = st.sidebar.file_uploader("Upload model allergen (alergen_model.pt)", type=["pt"]) if uploaded_model is not None: with open("alergen_model.pt", "wb") as f: f.write(uploaded_model.getbuffer()) st.sidebar.success("Model telah diupload dan dimuat!") # Load model model, tokenizer, target_columns = load_model() # Input area st.header("Masukkan Daftar Bahan Resep") ingredients = st.text_area("Bahan-bahan:", height=200, placeholder="Contoh: 1 bungkus Lontong homemade, 2 butir Telur ayam, 2 kotak kecil Tahu coklat...") col1, col2 = st.columns(2) with col1: if st.button("Deteksi Alergen", type="primary"): if ingredients: with st.spinner("Menganalisis bahan-bahan..."): # Clean text for display cleaned_text = clean_text(ingredients) st.markdown("### Bahan yang diproses:") st.text(cleaned_text) # Get predictions results = predict_allergens(ingredients, model, tokenizer, target_columns) # Display results st.markdown("### Hasil Deteksi Alergen:") # Create data for visualization allergens = list(results.keys()) probabilities = [results[a]['probability'] for a in allergens] present = [results[a]['present'] for a in allergens] # Create a colorful table of results result_df = pd.DataFrame({ 'Alergen': [a.title() for a in allergens], 'Terdeteksi': ['✅' if results[a]['present'] else '❌' for a in allergens], 'Probabilitas': [f"{results[a]['probability']*100:.2f}%" for a in allergens] }) st.dataframe(result_df, use_container_width=True) # Display chart in the second column with col2: fig, ax = plt.subplots(figsize=(10, 6)) bars = ax.bar( [a.title() for a in allergens], probabilities, color=['red' if p else 'green' for p in present] ) # Add threshold line ax.axhline(y=0.5, color='black', linestyle='--', alpha=0.7) ax.text(len(allergens)-1, 0.51, 'Threshold (0.5)', ha='right', va='bottom') # Customize the chart ax.set_ylim(0, 1) ax.set_ylabel('Probabilitas') ax.set_title('Probabilitas Deteksi Alergen') # Add values on top of bars for bar in bars: height = bar.get_height() ax.annotate(f'{height:.2f}', xy=(bar.get_x() + bar.get_width() / 2, height), xytext=(0, 3), # 3 points vertical offset textcoords="offset points", ha='center', va='bottom') st.pyplot(fig) # Show detailed explanation st.markdown("### Penjelasan Hasil:") detected_allergens = [allergen.title() for allergen, data in results.items() if data['present']] if detected_allergens: st.markdown(f"Resep ini kemungkinan mengandung alergen: **{', '.join(detected_allergens)}**") # Provide specific explanation for each detected allergen for allergen in detected_allergens: if allergen.lower() == 'susu': st.markdown("- **Susu**: Resep mungkin mengandung susu atau produk turunannya") elif allergen.lower() == 'kacang': st.markdown("- **Kacang**: Resep mungkin mengandung kacang atau produk turunannya") elif allergen.lower() == 'telur': st.markdown("- **Telur**: Resep mungkin mengandung telur atau produk turunannya") elif allergen.lower() == 'makanan_laut': st.markdown("- **Makanan Laut**: Resep mungkin mengandung ikan, udang, kerang, atau makanan laut lainnya") elif allergen.lower() == 'gandum': st.markdown("- **Gandum**: Resep mungkin mengandung gandum atau produk turunannya (termasuk gluten)") else: st.markdown("Tidak terdeteksi alergen umum dalam resep ini.") st.warning("Catatan: Prediksi ini hanya bersifat indikatif. Selalu verifikasi dengan informasi resmi untuk keamanan konsumsi.") else: st.error("Mohon masukkan daftar bahan terlebih dahulu.") # Examples section with st.expander("Contoh Resep"): st.markdown(""" ### Contoh Resep 1 (Mengandung Beberapa Alergen) ``` 1 bungkus Lontong homemade, 2 butir Telur ayam, 2 kotak kecil Tahu coklat, 4 butir kecil Kentang, 2 buah Tomat merah, 1 buah Ketimun lalap, 4 lembar Selada keriting, 2 lembar Kol putih, 2 porsi Saus kacang homemade, 4 buah Kerupuk udang goreng, Secukupnya emping goreng, 2 sdt Bawang goreng, Secukupnya Kecap manis ``` ### Contoh Resep 2 (Mengandung Susu) ``` 250 ml susu full cream, 2 sdm tepung maizena, 3 sdm gula pasir, 1/2 sdt vanila ekstrak, secukupnya keju cheddar parut ``` ### Contoh Resep 3 (Mengandung Makanan Laut) ``` 250 g udang segar, 150 g cumi-cumi, 2 sdm saus tiram, 3 siung bawang putih, 1 ruas jahe, 2 sdm minyak goreng, garam dan merica secukupnya ``` """) # About section st.sidebar.markdown("---") st.sidebar.header("Tentang") st.sidebar.info(""" Aplikasi ini menggunakan model deep learning berbasis IndoBERT untuk mendeteksi alergen dalam resep makanan. Model ini dilatih untuk mengidentifikasi 5 jenis alergen umum dalam makanan berdasarkan daftar bahan resep. """) # Model information st.sidebar.markdown("---") st.sidebar.header("Informasi Model") st.sidebar.markdown(""" - **Model Dasar**: IndoBERT - **Jenis**: Multilabel Classification - **Alergen yang Dideteksi**: Susu, Kacang, Telur, Makanan Laut, Gandum """) if __name__ == "__main__": main()