Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -11,6 +11,7 @@ from tensorflow.keras.applications.vgg16 import preprocess_input as vgg_preproce
|
|
11 |
from tensorflow.keras.applications.xception import preprocess_input as xce_preprocess
|
12 |
from tensorflow.keras.losses import BinaryFocalCrossentropy
|
13 |
from PIL import Image
|
|
|
14 |
|
15 |
# --- CONFIGURATION ---
|
16 |
FOREST_COORDS = {'Pakistan Forest': (34.0, 73.0)}
|
@@ -18,8 +19,8 @@ API_URL = (
|
|
18 |
"https://archive-api.open-meteo.com/v1/archive"
|
19 |
"?latitude={lat}&longitude={lon}"
|
20 |
"&start_date={start}&end_date={end}"
|
21 |
-
"&daily=temperature_2m_max,temperature_2m_min,"
|
22 |
-
"precipitation_sum,windspeed_10m_max,"
|
23 |
"relative_humidity_2m_max,relative_humidity_2m_min"
|
24 |
"&timezone=UTC"
|
25 |
)
|
@@ -44,15 +45,19 @@ def load_models():
|
|
44 |
'severity_post_tta.keras',
|
45 |
custom_objects={'focal_loss_fixed': focal_loss_fixed()}
|
46 |
)
|
|
|
|
|
|
|
47 |
rf_model = joblib.load('ensemble_rf_model.pkl')
|
48 |
-
xgb_model = joblib.load('ensemble_xgb_model.pkl')
|
49 |
lr_model = joblib.load('wildfire_logistic_model_synthetic.joblib')
|
50 |
return vgg_model, xce_model, rf_model, xgb_model, lr_model
|
51 |
except Exception as e:
|
52 |
print(f"Error loading models: {e}")
|
53 |
return None, None, None, None, None
|
54 |
|
55 |
-
#
|
|
|
|
|
56 |
target_map = {0: 'mild', 1: 'moderate', 2: 'severe'}
|
57 |
trend_map = {1: 'increase', 0: 'same', -1: 'decrease'}
|
58 |
task_rules = {
|
@@ -60,29 +65,7 @@ task_rules = {
|
|
60 |
'moderate':{'decrease':'mild','same':'moderate','increase':'severe'},
|
61 |
'severe': {'decrease':'moderate','same':'severe','increase':'severe'}
|
62 |
}
|
63 |
-
recommendations = {
|
64 |
-
'mild': {
|
65 |
-
'immediate': "Deploy spot crews for initial attack. Establish command post. Monitor fire behavior with drones or aircraft. Alert local fire stations.",
|
66 |
-
'evacuation': "No mass evacuation needed. Notify nearby communities of potential risk. Prepare evacuation routes if conditions change.",
|
67 |
-
'containment': "Establish initial fire lines. Use hand crews for direct attack. Position water resources. Clear fuel breaks where feasible.",
|
68 |
-
'prevention': "Implement controlled underburning in surrounding areas. Manage vegetation density. Create defensible spaces around structures.",
|
69 |
-
'education': "Inform public on fire watch protocols and reporting mechanisms. Train local volunteers in basic firefighting techniques."
|
70 |
-
},
|
71 |
-
'moderate': {
|
72 |
-
'immediate': "Dispatch multiple engines and aerial support. Establish unified command system. Deploy heavy equipment. Request additional resources.",
|
73 |
-
'evacuation': "Prepare evacuation zones and staging areas. Advise voluntary evacuation for vulnerable populations. Alert emergency shelters.",
|
74 |
-
'containment': "Build substantial fire breaks. Conduct water drops from helicopters. Implement indirect attack strategies. Protect critical infrastructure.",
|
75 |
-
'prevention': "Initiate fuel reduction programs in adjacent areas. Create wider buffer zones. Assess watershed protection needs.",
|
76 |
-
'education': "Conduct community emergency drills. Launch awareness campaigns on evacuation procedures. Distribute preparedness materials."
|
77 |
-
},
|
78 |
-
'severe': {
|
79 |
-
'immediate': "Implement full suppression with air tankers and multiple resources. Establish incident management team. Request state/federal assistance. Deploy specialized teams.",
|
80 |
-
'evacuation': "Issue mandatory evacuation orders. Open multiple emergency shelters. Implement traffic control measures. Assist vulnerable populations.",
|
81 |
-
'containment': "Deploy fire retardant lines from aircraft. Consider backfires and burnout operations. Protect critical infrastructure. Establish multiple control lines.",
|
82 |
-
'prevention': "Plan for reforestation and erosion control. Harden infrastructure against future fires. Implement watershed protection measures.",
|
83 |
-
'education': "Conduct comprehensive emergency response training. Implement risk communication strategies. Develop long-term community resilience programs."
|
84 |
-
}
|
85 |
-
}
|
86 |
|
87 |
# --- PIPELINE FUNCTIONS ---
|
88 |
def detect_fire(img):
|
@@ -123,7 +106,6 @@ def fetch_weather_trend(lat, lon):
|
|
123 |
response.raise_for_status()
|
124 |
df = pd.DataFrame(response.json().get('daily', {}))
|
125 |
except Exception:
|
126 |
-
# Fallback sample data
|
127 |
df = pd.DataFrame({
|
128 |
'date': [(datetime.utcnow() - timedelta(days=i)).strftime('%Y-%m-%d') for i in range(1,-1,-1)],
|
129 |
'precipitation_sum': [5, 2],
|
@@ -133,15 +115,13 @@ def fetch_weather_trend(lat, lon):
|
|
133 |
'relative_humidity_2m_min': [40, 35],
|
134 |
'windspeed_10m_max': [15, 18]
|
135 |
})
|
136 |
-
# Numeric conversions
|
137 |
for c in ['precipitation_sum','temperature_2m_max','temperature_2m_min',
|
138 |
'relative_humidity_2m_max','relative_humidity_2m_min','windspeed_10m_max']:
|
139 |
df[c] = pd.to_numeric(df[c], errors='coerce')
|
140 |
-
# Feature engineering
|
141 |
-
df['precipitation'] = df['precipitation_sum']
|
142 |
df['temperature'] = (df['temperature_2m_max'] + df['temperature_2m_min']) / 2
|
143 |
df['humidity'] = (df['relative_humidity_2m_max'] + df['relative_humidity_2m_min']) / 2
|
144 |
df['wind_speed'] = df['windspeed_10m_max']
|
|
|
145 |
df['fire_risk_score'] = (
|
146 |
0.4 * (df['temperature'] / 55) +
|
147 |
0.2 * (1 - df['humidity'] / 100) +
|
@@ -157,19 +137,16 @@ def fetch_weather_trend(lat, lon):
|
|
157 |
def generate_recommendations(original_severity, weather_trend):
|
158 |
projected = task_rules[original_severity][weather_trend]
|
159 |
rec = recommendations[projected]
|
160 |
-
return (f"**Original Severity:** {original_severity.title()} \
|
161 |
-
" \
|
162 |
-
|
163 |
-
" \
|
164 |
-
|
165 |
-
|
166 |
-
|
167 |
-
|
168 |
-
|
169 |
-
f"**Prevention:** {rec['prevention']}\n\n" \
|
170 |
-
f"**Education:** {rec['education']}")
|
171 |
|
172 |
-
# --- MAIN PIPELINE ---
|
173 |
def pipeline(image):
|
174 |
if image is None:
|
175 |
return "No image provided","N/A","N/A","**Please upload an image to analyze**"
|
@@ -191,7 +168,6 @@ def pipeline(image):
|
|
191 |
recs
|
192 |
)
|
193 |
|
194 |
-
# --- SAFE WRAPPER FOR UI ---
|
195 |
def safe_pipeline(image):
|
196 |
try:
|
197 |
return pipeline(image)
|
@@ -199,32 +175,19 @@ def safe_pipeline(image):
|
|
199 |
print(f"Error in pipeline: {e}")
|
200 |
return "Error during analysis","N/A","N/A", f"**Error: {e}**"
|
201 |
|
202 |
-
# ---
|
203 |
-
vgg_model, xce_model, rf_model, xgb_model, lr_model = load_models()
|
204 |
-
|
205 |
-
# --- UI: CUSTOM CSS & GRADIO LAYOUT ---
|
206 |
custom_css = '''
|
207 |
-
#header { text-align: center; margin:
|
208 |
-
#header img { height: 4rem; margin-right: 1rem; }
|
209 |
-
#main-title { font-size: 2.75rem; margin: 0.5rem 0; }
|
210 |
-
#sub-title { font-size: 1.25rem; color: #555; }
|
211 |
-
.gr-button.primary { background: #ff7043 !important; }
|
212 |
-
.output-card { background: #f7f7f7; border-radius: 0.75rem; padding: 1rem;
|
213 |
-
box-shadow: 0 1px 6px rgba(0,0,0,0.1); margin-bottom: 1rem; }
|
214 |
'''
|
215 |
-
|
216 |
with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
|
217 |
-
# Header (add your logo.png in working directory or adjust path)
|
218 |
with gr.Row(elem_id="header"):
|
219 |
try:
|
220 |
gr.Image(value="logo.png", show_label=False)
|
221 |
except:
|
222 |
pass
|
223 |
with gr.Column():
|
224 |
-
gr.Markdown("# 🔥 Wildfire Command Center"
|
225 |
-
gr.Markdown("Upload a forest image to detect wildfire, classify severity, and get actionable recommendations."
|
226 |
-
|
227 |
-
# Tabs: Analyze & Last Analysis
|
228 |
with gr.Tabs():
|
229 |
with gr.TabItem("Analyze 🔍"):
|
230 |
with gr.Row():
|
@@ -232,19 +195,15 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
|
|
232 |
image_input = gr.Image(type="numpy", label="Forest Image", tool="editor")
|
233 |
run_btn = gr.Button("Analyze Now", variant="primary")
|
234 |
with gr.Column(scale=1):
|
235 |
-
|
236 |
-
|
237 |
-
|
238 |
-
|
239 |
-
recs_out = gr.Markdown("---", label="Recommendations")
|
240 |
-
|
241 |
with gr.TabItem("Last Analysis 📊"):
|
242 |
last_status = gr.Markdown("*No analysis yet*", elem_classes="output-card")
|
243 |
last_severity = gr.Markdown("---", elem_classes="output-card")
|
244 |
last_trend = gr.Markdown("---", elem_classes="output-card")
|
245 |
last_recs = gr.Markdown("---", elem_classes="output-card")
|
246 |
-
|
247 |
-
# Bind actions: analyze then archive outputs
|
248 |
run_btn.click(
|
249 |
fn=safe_pipeline,
|
250 |
inputs=image_input,
|
@@ -254,6 +213,5 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
|
|
254 |
inputs=[status_out, severity_out, trend_out, recs_out],
|
255 |
outputs=[last_status, last_severity, last_trend, last_recs]
|
256 |
)
|
257 |
-
|
258 |
if __name__ == '__main__':
|
259 |
demo.queue(api_open=True).launch()
|
|
|
11 |
from tensorflow.keras.applications.xception import preprocess_input as xce_preprocess
|
12 |
from tensorflow.keras.losses import BinaryFocalCrossentropy
|
13 |
from PIL import Image
|
14 |
+
from xgboost import XGBClassifier
|
15 |
|
16 |
# --- CONFIGURATION ---
|
17 |
FOREST_COORDS = {'Pakistan Forest': (34.0, 73.0)}
|
|
|
19 |
"https://archive-api.open-meteo.com/v1/archive"
|
20 |
"?latitude={lat}&longitude={lon}"
|
21 |
"&start_date={start}&end_date={end}"
|
22 |
+
"&daily=temperature_2m_max,temperature_2m_min,"
|
23 |
+
"precipitation_sum,windspeed_10m_max,"
|
24 |
"relative_humidity_2m_max,relative_humidity_2m_min"
|
25 |
"&timezone=UTC"
|
26 |
)
|
|
|
45 |
'severity_post_tta.keras',
|
46 |
custom_objects={'focal_loss_fixed': focal_loss_fixed()}
|
47 |
)
|
48 |
+
# Reload XGBoost from JSON to avoid pickle warnings
|
49 |
+
xgb_model = XGBClassifier()
|
50 |
+
xgb_model.load_model('ensemble_xgb_model.json')
|
51 |
rf_model = joblib.load('ensemble_rf_model.pkl')
|
|
|
52 |
lr_model = joblib.load('wildfire_logistic_model_synthetic.joblib')
|
53 |
return vgg_model, xce_model, rf_model, xgb_model, lr_model
|
54 |
except Exception as e:
|
55 |
print(f"Error loading models: {e}")
|
56 |
return None, None, None, None, None
|
57 |
|
58 |
+
# Load models once
|
59 |
+
vgg_model, xce_model, rf_model, xgb_model, lr_model = load_models()
|
60 |
+
|
61 |
target_map = {0: 'mild', 1: 'moderate', 2: 'severe'}
|
62 |
trend_map = {1: 'increase', 0: 'same', -1: 'decrease'}
|
63 |
task_rules = {
|
|
|
65 |
'moderate':{'decrease':'mild','same':'moderate','increase':'severe'},
|
66 |
'severe': {'decrease':'moderate','same':'severe','increase':'severe'}
|
67 |
}
|
68 |
+
recommendations = { ... } # (keep your existing recommendations dict here)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
|
70 |
# --- PIPELINE FUNCTIONS ---
|
71 |
def detect_fire(img):
|
|
|
106 |
response.raise_for_status()
|
107 |
df = pd.DataFrame(response.json().get('daily', {}))
|
108 |
except Exception:
|
|
|
109 |
df = pd.DataFrame({
|
110 |
'date': [(datetime.utcnow() - timedelta(days=i)).strftime('%Y-%m-%d') for i in range(1,-1,-1)],
|
111 |
'precipitation_sum': [5, 2],
|
|
|
115 |
'relative_humidity_2m_min': [40, 35],
|
116 |
'windspeed_10m_max': [15, 18]
|
117 |
})
|
|
|
118 |
for c in ['precipitation_sum','temperature_2m_max','temperature_2m_min',
|
119 |
'relative_humidity_2m_max','relative_humidity_2m_min','windspeed_10m_max']:
|
120 |
df[c] = pd.to_numeric(df[c], errors='coerce')
|
|
|
|
|
121 |
df['temperature'] = (df['temperature_2m_max'] + df['temperature_2m_min']) / 2
|
122 |
df['humidity'] = (df['relative_humidity_2m_max'] + df['relative_humidity_2m_min']) / 2
|
123 |
df['wind_speed'] = df['windspeed_10m_max']
|
124 |
+
df['precipitation'] = df['precipitation_sum']
|
125 |
df['fire_risk_score'] = (
|
126 |
0.4 * (df['temperature'] / 55) +
|
127 |
0.2 * (1 - df['humidity'] / 100) +
|
|
|
137 |
def generate_recommendations(original_severity, weather_trend):
|
138 |
projected = task_rules[original_severity][weather_trend]
|
139 |
rec = recommendations[projected]
|
140 |
+
return (f"**Original Severity:** {original_severity.title()} \" \
|
141 |
+
f"**Weather Trend:** {weather_trend.title()} \" \
|
142 |
+
f"**Projected Severity:** {projected.title()}\n\n" \
|
143 |
+
"### Management Recommendations:\n" \
|
144 |
+
f"**Immediate:** {rec['immediate']}\n\n" \
|
145 |
+
f"**Evacuation:** {rec['evacuation']}\n\n" \
|
146 |
+
f"**Containment:** {rec['containment']}\n\n" \
|
147 |
+
f"**Prevention:** {rec['prevention']}\n\n" \
|
148 |
+
f"**Education:** {rec['education']}")
|
|
|
|
|
149 |
|
|
|
150 |
def pipeline(image):
|
151 |
if image is None:
|
152 |
return "No image provided","N/A","N/A","**Please upload an image to analyze**"
|
|
|
168 |
recs
|
169 |
)
|
170 |
|
|
|
171 |
def safe_pipeline(image):
|
172 |
try:
|
173 |
return pipeline(image)
|
|
|
175 |
print(f"Error in pipeline: {e}")
|
176 |
return "Error during analysis","N/A","N/A", f"**Error: {e}**"
|
177 |
|
178 |
+
# --- GRADIO UI ---
|
|
|
|
|
|
|
179 |
custom_css = '''
|
180 |
+
#header { text-align: center; margin-bottom: 1rem; }
|
|
|
|
|
|
|
|
|
|
|
|
|
181 |
'''
|
|
|
182 |
with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
|
|
|
183 |
with gr.Row(elem_id="header"):
|
184 |
try:
|
185 |
gr.Image(value="logo.png", show_label=False)
|
186 |
except:
|
187 |
pass
|
188 |
with gr.Column():
|
189 |
+
gr.Markdown("# 🔥 Wildfire Command Center")
|
190 |
+
gr.Markdown("Upload a forest image to detect wildfire, classify severity, and get actionable recommendations.")
|
|
|
|
|
191 |
with gr.Tabs():
|
192 |
with gr.TabItem("Analyze 🔍"):
|
193 |
with gr.Row():
|
|
|
195 |
image_input = gr.Image(type="numpy", label="Forest Image", tool="editor")
|
196 |
run_btn = gr.Button("Analyze Now", variant="primary")
|
197 |
with gr.Column(scale=1):
|
198 |
+
status_out = gr.Markdown("*Status will appear here*", label="Status")
|
199 |
+
severity_out = gr.Markdown("---", label="Severity")
|
200 |
+
trend_out = gr.Markdown("---", label="Weather Trend")
|
201 |
+
recs_out = gr.Markdown("---", label="Recommendations")
|
|
|
|
|
202 |
with gr.TabItem("Last Analysis 📊"):
|
203 |
last_status = gr.Markdown("*No analysis yet*", elem_classes="output-card")
|
204 |
last_severity = gr.Markdown("---", elem_classes="output-card")
|
205 |
last_trend = gr.Markdown("---", elem_classes="output-card")
|
206 |
last_recs = gr.Markdown("---", elem_classes="output-card")
|
|
|
|
|
207 |
run_btn.click(
|
208 |
fn=safe_pipeline,
|
209 |
inputs=image_input,
|
|
|
213 |
inputs=[status_out, severity_out, trend_out, recs_out],
|
214 |
outputs=[last_status, last_severity, last_trend, last_recs]
|
215 |
)
|
|
|
216 |
if __name__ == '__main__':
|
217 |
demo.queue(api_open=True).launch()
|