Spaces:
Running
Running
rdsarjito
commited on
Commit
·
c0cfde6
1
Parent(s):
552cd20
2 commit
Browse files- app.py +171 -186
- requirements.txt +4 -7
app.py
CHANGED
@@ -1,27 +1,40 @@
|
|
1 |
import streamlit as st
|
2 |
-
import os
|
3 |
-
import numpy as np
|
4 |
-
import pandas as pd
|
5 |
-
import re
|
6 |
import torch
|
7 |
import torch.nn as nn
|
8 |
-
|
9 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
10 |
-
import
|
11 |
-
import
|
12 |
-
warnings.filterwarnings("ignore")
|
13 |
|
14 |
# Set page config
|
15 |
st.set_page_config(
|
16 |
-
page_title="Deteksi Alergen
|
17 |
-
page_icon="
|
18 |
layout="wide"
|
19 |
)
|
20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
# Set device
|
22 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
23 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
# Clean text function
|
|
|
25 |
def clean_text(text):
|
26 |
# Convert dashes to spaces for better tokenization
|
27 |
text = text.replace('--', ' ')
|
@@ -46,36 +59,38 @@ class MultilabelBertClassifier(nn.Module):
|
|
46 |
outputs = self.bert(input_ids=input_ids, attention_mask=attention_mask)
|
47 |
return outputs.logits
|
48 |
|
49 |
-
# Function to predict allergens in new recipes
|
50 |
@st.cache_resource
|
51 |
-
def
|
52 |
-
# Target columns
|
53 |
-
target_columns = ['susu', 'kacang', 'telur', 'makanan_laut', 'gandum']
|
54 |
-
|
55 |
-
# Initialize tokenizer
|
56 |
-
tokenizer = AutoTokenizer.from_pretrained('indobenchmark/indobert-base-p2')
|
57 |
-
|
58 |
-
# Initialize model
|
59 |
-
model = MultilabelBertClassifier('indobenchmark/indobert-base-p1', len(target_columns))
|
60 |
-
|
61 |
-
# Load model weights if available
|
62 |
-
model_path = "model/alergen_model.pt"
|
63 |
-
|
64 |
try:
|
65 |
-
#
|
66 |
-
|
67 |
-
|
68 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
except Exception as e:
|
70 |
-
st.error(f"Error loading model: {
|
71 |
-
|
72 |
-
|
73 |
-
model.to(device)
|
74 |
-
model.eval()
|
75 |
-
|
76 |
-
return model, tokenizer, target_columns
|
77 |
|
78 |
-
|
|
|
|
|
|
|
|
|
79 |
# Clean the text
|
80 |
cleaned_text = clean_text(ingredients_text)
|
81 |
|
@@ -95,170 +110,140 @@ def predict_allergens(ingredients_text, model, tokenizer, target_columns, max_le
|
|
95 |
with torch.no_grad():
|
96 |
outputs = model(input_ids=input_ids, attention_mask=attention_mask)
|
97 |
predictions = torch.sigmoid(outputs)
|
98 |
-
|
99 |
-
|
100 |
|
101 |
result = {}
|
|
|
102 |
for i, target in enumerate(target_columns):
|
103 |
-
result[target] =
|
104 |
-
|
105 |
-
'probability': float(predictions_prob[i])
|
106 |
-
}
|
107 |
|
108 |
-
return result
|
109 |
|
110 |
-
#
|
111 |
-
|
112 |
-
st.
|
113 |
-
st.
|
114 |
-
Aplikasi ini menggunakan model IndoBERT untuk mendeteksi kemungkinan alergen dalam resep berdasarkan daftar bahan.
|
115 |
-
Alergen yang diidentifikasi meliputi:
|
116 |
-
- Susu
|
117 |
-
- Kacang
|
118 |
-
- Telur
|
119 |
-
- Makanan Laut
|
120 |
-
- Gandum
|
121 |
-
""")
|
122 |
-
|
123 |
-
# Sidebar for model upload
|
124 |
-
st.sidebar.header("Upload Model")
|
125 |
-
uploaded_model = st.sidebar.file_uploader("Upload model allergen (alergen_model.pt)", type=["pt"])
|
126 |
|
127 |
if uploaded_model is not None:
|
128 |
with open("alergen_model.pt", "wb") as f:
|
129 |
f.write(uploaded_model.getbuffer())
|
130 |
-
st.
|
|
|
131 |
|
132 |
-
|
133 |
-
|
134 |
-
|
135 |
-
|
136 |
-
|
137 |
-
|
138 |
-
|
139 |
-
|
140 |
-
|
141 |
-
|
142 |
-
|
143 |
-
|
144 |
-
|
145 |
-
|
146 |
-
|
147 |
-
|
148 |
-
|
149 |
-
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
|
154 |
-
|
155 |
-
|
156 |
-
|
157 |
-
|
158 |
-
|
159 |
-
|
160 |
-
|
161 |
-
|
162 |
-
|
163 |
-
|
164 |
-
|
165 |
-
|
166 |
-
|
167 |
-
|
168 |
-
|
169 |
-
|
170 |
-
|
171 |
-
# Display chart in the second column
|
172 |
-
with col2:
|
173 |
-
fig, ax = plt.subplots(figsize=(10, 6))
|
174 |
-
bars = ax.bar(
|
175 |
-
[a.title() for a in allergens],
|
176 |
-
probabilities,
|
177 |
-
color=['red' if p else 'green' for p in present]
|
178 |
-
)
|
179 |
-
|
180 |
-
# Add threshold line
|
181 |
-
ax.axhline(y=0.5, color='black', linestyle='--', alpha=0.7)
|
182 |
-
ax.text(len(allergens)-1, 0.51, 'Threshold (0.5)', ha='right', va='bottom')
|
183 |
-
|
184 |
-
# Customize the chart
|
185 |
-
ax.set_ylim(0, 1)
|
186 |
-
ax.set_ylabel('Probabilitas')
|
187 |
-
ax.set_title('Probabilitas Deteksi Alergen')
|
188 |
-
|
189 |
-
# Add values on top of bars
|
190 |
-
for bar in bars:
|
191 |
-
height = bar.get_height()
|
192 |
-
ax.annotate(f'{height:.2f}',
|
193 |
-
xy=(bar.get_x() + bar.get_width() / 2, height),
|
194 |
-
xytext=(0, 3), # 3 points vertical offset
|
195 |
-
textcoords="offset points",
|
196 |
-
ha='center', va='bottom')
|
197 |
-
|
198 |
-
st.pyplot(fig)
|
199 |
-
|
200 |
-
# Show detailed explanation
|
201 |
-
st.markdown("### Penjelasan Hasil:")
|
202 |
-
detected_allergens = [allergen.title() for allergen, data in results.items() if data['present']]
|
203 |
|
204 |
-
|
205 |
-
|
206 |
-
|
207 |
-
# Provide specific explanation for each detected allergen
|
208 |
-
for allergen in detected_allergens:
|
209 |
-
if allergen.lower() == 'susu':
|
210 |
-
st.markdown("- **Susu**: Resep mungkin mengandung susu atau produk turunannya")
|
211 |
-
elif allergen.lower() == 'kacang':
|
212 |
-
st.markdown("- **Kacang**: Resep mungkin mengandung kacang atau produk turunannya")
|
213 |
-
elif allergen.lower() == 'telur':
|
214 |
-
st.markdown("- **Telur**: Resep mungkin mengandung telur atau produk turunannya")
|
215 |
-
elif allergen.lower() == 'makanan_laut':
|
216 |
-
st.markdown("- **Makanan Laut**: Resep mungkin mengandung ikan, udang, kerang, atau makanan laut lainnya")
|
217 |
-
elif allergen.lower() == 'gandum':
|
218 |
-
st.markdown("- **Gandum**: Resep mungkin mengandung gandum atau produk turunannya (termasuk gluten)")
|
219 |
-
else:
|
220 |
-
st.markdown("Tidak terdeteksi alergen umum dalam resep ini.")
|
221 |
|
222 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
223 |
else:
|
224 |
-
st.error("
|
225 |
-
|
226 |
-
# Examples section
|
227 |
-
with st.expander("Contoh Resep"):
|
228 |
-
st.markdown("""
|
229 |
-
### Contoh Resep 1 (Mengandung Beberapa Alergen)
|
230 |
-
```
|
231 |
-
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
|
232 |
-
```
|
233 |
-
|
234 |
-
### Contoh Resep 2 (Mengandung Susu)
|
235 |
-
```
|
236 |
-
250 ml susu full cream, 2 sdm tepung maizena, 3 sdm gula pasir, 1/2 sdt vanila ekstrak, secukupnya keju cheddar parut
|
237 |
-
```
|
238 |
-
|
239 |
-
### Contoh Resep 3 (Mengandung Makanan Laut)
|
240 |
-
```
|
241 |
-
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
|
242 |
-
```
|
243 |
-
""")
|
244 |
|
245 |
-
|
246 |
-
|
247 |
-
st.
|
248 |
-
|
249 |
-
|
250 |
-
|
251 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
252 |
""")
|
253 |
|
254 |
-
|
255 |
-
|
256 |
-
|
257 |
-
|
258 |
-
|
259 |
-
|
260 |
-
|
261 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
262 |
|
263 |
-
|
264 |
-
|
|
|
|
1 |
import streamlit as st
|
|
|
|
|
|
|
|
|
2 |
import torch
|
3 |
import torch.nn as nn
|
4 |
+
import re
|
5 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
6 |
+
import os
|
7 |
+
import numpy as np
|
|
|
8 |
|
9 |
# Set page config
|
10 |
st.set_page_config(
|
11 |
+
page_title="Deteksi Alergen Resep",
|
12 |
+
page_icon="🍽️",
|
13 |
layout="wide"
|
14 |
)
|
15 |
|
16 |
+
# App title and description
|
17 |
+
st.title("🍽️ Deteksi Alergen Resep Makanan")
|
18 |
+
st.markdown("""
|
19 |
+
Aplikasi ini dapat mendeteksi potensi alergen dalam resep makanan Indonesia.
|
20 |
+
Masukkan daftar bahan-bahan resep Anda, dan sistem akan mengidentifikasi alergen yang mungkin terkandung.
|
21 |
+
""")
|
22 |
+
|
23 |
# Set device
|
24 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
25 |
|
26 |
+
# Define target columns (allergens)
|
27 |
+
target_columns = ['susu', 'kacang', 'telur', 'makanan_laut', 'gandum']
|
28 |
+
allergen_descriptions = {
|
29 |
+
'susu': 'Produk susu (milk products)',
|
30 |
+
'kacang': 'Kacang-kacangan (nuts)',
|
31 |
+
'telur': 'Telur (eggs)',
|
32 |
+
'makanan_laut': 'Makanan laut (seafood)',
|
33 |
+
'gandum': 'Gandum/gluten (wheat/gluten)'
|
34 |
+
}
|
35 |
+
|
36 |
# Clean text function
|
37 |
+
@st.cache_data
|
38 |
def clean_text(text):
|
39 |
# Convert dashes to spaces for better tokenization
|
40 |
text = text.replace('--', ' ')
|
|
|
59 |
outputs = self.bert(input_ids=input_ids, attention_mask=attention_mask)
|
60 |
return outputs.logits
|
61 |
|
|
|
62 |
@st.cache_resource
|
63 |
+
def load_model_and_tokenizer():
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
64 |
try:
|
65 |
+
# Initialize tokenizer
|
66 |
+
tokenizer = AutoTokenizer.from_pretrained('indobenchmark/indobert-base-p2')
|
67 |
+
|
68 |
+
# Initialize model
|
69 |
+
model = MultilabelBertClassifier('indobenchmark/indobert-base-p1', len(target_columns))
|
70 |
+
|
71 |
+
# Check if model exists locally, otherwise download from Hugging Face
|
72 |
+
model_path = "alergen_model.pt"
|
73 |
+
|
74 |
+
if os.path.exists(model_path):
|
75 |
+
st.info("Loading model from local storage...")
|
76 |
+
checkpoint = torch.load(model_path, map_location=device)
|
77 |
+
model.load_state_dict(checkpoint['model_state_dict'])
|
78 |
+
else:
|
79 |
+
st.warning("Model file not found. Please upload your model file.")
|
80 |
+
|
81 |
+
model.to(device)
|
82 |
+
model.eval()
|
83 |
+
|
84 |
+
return model, tokenizer
|
85 |
except Exception as e:
|
86 |
+
st.error(f"Error loading model: {e}")
|
87 |
+
return None, None
|
|
|
|
|
|
|
|
|
|
|
88 |
|
89 |
+
# Function to predict allergens in new recipes
|
90 |
+
def predict_allergens(model, tokenizer, ingredients_text, max_length=128):
|
91 |
+
if not model or not tokenizer:
|
92 |
+
return None
|
93 |
+
|
94 |
# Clean the text
|
95 |
cleaned_text = clean_text(ingredients_text)
|
96 |
|
|
|
110 |
with torch.no_grad():
|
111 |
outputs = model(input_ids=input_ids, attention_mask=attention_mask)
|
112 |
predictions = torch.sigmoid(outputs)
|
113 |
+
predictions_np = predictions.cpu().numpy()[0]
|
114 |
+
binary_predictions = (predictions > 0.5).float().cpu().numpy()[0]
|
115 |
|
116 |
result = {}
|
117 |
+
confidence = {}
|
118 |
for i, target in enumerate(target_columns):
|
119 |
+
result[target] = bool(binary_predictions[i])
|
120 |
+
confidence[target] = float(predictions_np[i])
|
|
|
|
|
121 |
|
122 |
+
return result, confidence
|
123 |
|
124 |
+
# Sidebar for model upload
|
125 |
+
with st.sidebar:
|
126 |
+
st.header("Model Management")
|
127 |
+
uploaded_model = st.file_uploader("Upload model file (alergen_model.pt)", type=["pt"])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
128 |
|
129 |
if uploaded_model is not None:
|
130 |
with open("alergen_model.pt", "wb") as f:
|
131 |
f.write(uploaded_model.getbuffer())
|
132 |
+
st.success("Model uploaded successfully!")
|
133 |
+
st.cache_resource.clear()
|
134 |
|
135 |
+
st.markdown("---")
|
136 |
+
st.markdown("### Tentang Aplikasi")
|
137 |
+
st.markdown("""
|
138 |
+
Aplikasi ini menggunakan model deep learning berbasis IndoBERT untuk mendeteksi
|
139 |
+
potensi alergen dalam resep makanan. Model dilatih untuk mendeteksi lima jenis alergen
|
140 |
+
umum dalam makanan.
|
141 |
+
""")
|
142 |
+
|
143 |
+
# Load model and tokenizer
|
144 |
+
model, tokenizer = load_model_and_tokenizer()
|
145 |
+
|
146 |
+
# Main content
|
147 |
+
st.header("Masukkan Bahan-bahan Resep")
|
148 |
+
|
149 |
+
# Text area for ingredients input
|
150 |
+
ingredients = st.text_area(
|
151 |
+
"Daftar Bahan (satu per baris atau dengan format yang umum digunakan)",
|
152 |
+
height=150,
|
153 |
+
placeholder="Contoh:\n1 bungkus Lontong homemade\n2 butir Telur ayam\n2 kotak kecil Tahu coklat\n4 butir kecil Kentang\n..."
|
154 |
+
)
|
155 |
+
|
156 |
+
# Predict button
|
157 |
+
if st.button("Deteksi Alergen", type="primary"):
|
158 |
+
if not ingredients:
|
159 |
+
st.warning("Silakan masukkan daftar bahan terlebih dahulu.")
|
160 |
+
elif not model:
|
161 |
+
st.error("Model belum tersedia. Silakan upload model terlebih dahulu.")
|
162 |
+
else:
|
163 |
+
with st.spinner("Menganalisis resep..."):
|
164 |
+
results, confidence = predict_allergens(model, tokenizer, ingredients)
|
165 |
+
|
166 |
+
if results:
|
167 |
+
st.header("Hasil Deteksi Alergen")
|
168 |
+
|
169 |
+
# Display detected allergens
|
170 |
+
detected_allergens = [allergen for allergen, present in results.items() if present]
|
171 |
+
|
172 |
+
if detected_allergens:
|
173 |
+
st.markdown("### ⚠️ Alergen Terdeteksi:")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
174 |
|
175 |
+
# Create columns for the allergen cards
|
176 |
+
cols = st.columns(len(detected_allergens) if len(detected_allergens) < 3 else 3)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
177 |
|
178 |
+
for i, allergen in enumerate(detected_allergens):
|
179 |
+
col_idx = i % 3
|
180 |
+
with cols[col_idx]:
|
181 |
+
st.markdown(f"""
|
182 |
+
<div style="padding: 10px; border-radius: 5px; background-color: #ffeeee; margin-bottom: 10px;">
|
183 |
+
<h4 style="color: #cc0000;">{allergen_descriptions[allergen]}</h4>
|
184 |
+
<p>Tingkat kepercayaan: {confidence[allergen]*100:.1f}%</p>
|
185 |
+
</div>
|
186 |
+
""", unsafe_allow_html=True)
|
187 |
+
else:
|
188 |
+
st.success("✅ Tidak ada alergen yang terdeteksi dalam resep ini.")
|
189 |
+
|
190 |
+
# Display detailed analysis
|
191 |
+
with st.expander("Lihat Analisis Detail"):
|
192 |
+
st.markdown("### Tingkat Kepercayaan Per Alergen")
|
193 |
+
for allergen in target_columns:
|
194 |
+
conf_value = confidence[allergen]
|
195 |
+
st.markdown(f"**{allergen_descriptions[allergen]}:** {conf_value*100:.1f}%")
|
196 |
+
st.progress(conf_value)
|
197 |
else:
|
198 |
+
st.error("Terjadi kesalahan dalam prediksi. Silakan coba lagi.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
199 |
|
200 |
+
# Example recipe section
|
201 |
+
with st.expander("Lihat Contoh Resep"):
|
202 |
+
st.markdown("""
|
203 |
+
**Gado-gado:**
|
204 |
+
1 bungkus Lontong homemade
|
205 |
+
2 butir Telur ayam
|
206 |
+
2 kotak kecil Tahu coklat
|
207 |
+
4 butir kecil Kentang
|
208 |
+
2 buah Tomat merah
|
209 |
+
1 buah Ketimun lalap
|
210 |
+
4 lembar Selada keriting
|
211 |
+
2 lembar Kol putih
|
212 |
+
2 porsi Saus kacang homemade
|
213 |
+
4 buah Kerupuk udang goreng
|
214 |
+
Secukupnya emping goreng
|
215 |
+
2 sdt Bawang goreng
|
216 |
+
Secukupnya Kecap manis
|
217 |
""")
|
218 |
|
219 |
+
if st.button("Gunakan Contoh Ini"):
|
220 |
+
st.session_state.example_used = True
|
221 |
+
# Will be processed in next rerun
|
222 |
+
|
223 |
+
# Handle example
|
224 |
+
if 'example_used' in st.session_state and st.session_state.example_used:
|
225 |
+
example_recipe = """1 bungkus Lontong homemade
|
226 |
+
2 butir Telur ayam
|
227 |
+
2 kotak kecil Tahu coklat
|
228 |
+
4 butir kecil Kentang
|
229 |
+
2 buah Tomat merah
|
230 |
+
1 buah Ketimun lalap
|
231 |
+
4 lembar Selada keriting
|
232 |
+
2 lembar Kol putih
|
233 |
+
2 porsi Saus kacang homemade
|
234 |
+
4 buah Kerupuk udang goreng
|
235 |
+
Secukupnya emping goreng
|
236 |
+
2 sdt Bawang goreng
|
237 |
+
Secukupnya Kecap manis"""
|
238 |
+
|
239 |
+
st.session_state.example_used = False
|
240 |
+
st.text_area(
|
241 |
+
"Daftar Bahan (satu per baris atau dengan format yang umum digunakan)",
|
242 |
+
value=example_recipe,
|
243 |
+
height=150,
|
244 |
+
key="ingredients_example"
|
245 |
+
)
|
246 |
|
247 |
+
# Footer
|
248 |
+
st.markdown("---")
|
249 |
+
st.markdown("*Aplikasi ini hanya untuk tujuan informasi. Silakan konsultasikan dengan ahli gizi untuk konfirmasi alergen dalam makanan.*")
|
requirements.txt
CHANGED
@@ -1,8 +1,5 @@
|
|
1 |
-
streamlit>=1.
|
2 |
torch>=2.0.0
|
3 |
-
transformers>=4.
|
4 |
-
|
5 |
-
|
6 |
-
matplotlib>=3.7.0
|
7 |
-
scikit-learn>=1.3.0
|
8 |
-
regex>=20
|
|
|
1 |
+
streamlit>=1.24.0
|
2 |
torch>=2.0.0
|
3 |
+
transformers>=4.30.0
|
4 |
+
numpy>=1.22.0
|
5 |
+
regex>=2022.1.18
|
|
|
|
|
|