|
|
|
|
|
|
|
|
|
|
|
import os
|
|
|
import json
|
|
|
import ee
|
|
|
import geemap
|
|
|
import gradio as gr
|
|
|
import html
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
print("\n--- Инициализация Google Earth Engine ---")
|
|
|
try:
|
|
|
|
|
|
if 'GEE_CREDENTIALS' in os.environ:
|
|
|
print("Аутентификация через секрет Hugging Face...")
|
|
|
|
|
|
creds_json_str = os.environ['GEE_CREDENTIALS']
|
|
|
creds_json = json.loads(creds_json_str)
|
|
|
|
|
|
|
|
|
credentials = ee.ServiceAccountCredentials(creds_json['client_email'], key_data=creds_json_str)
|
|
|
ee.Initialize(credentials=credentials, project=creds_json['project_id'])
|
|
|
print("✅ Аутентификация через сервисный аккаунт GEE прошла успешно.")
|
|
|
else:
|
|
|
|
|
|
print("Секрет не найден, попытка локальной аутентификации...")
|
|
|
ee.Authenticate()
|
|
|
ee.Initialize(project='gen-lang-client-0605302377')
|
|
|
print("✅ Локальная аутентификация и инициализация GEE прошли успешно.")
|
|
|
|
|
|
except Exception as e:
|
|
|
print(f"🔴 Критическая ошибка на этапе инициализации GEE: {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
gee_classifier = None
|
|
|
bands_for_training = ['B2', 'B3', 'B4', 'B8', 'NDVI']
|
|
|
|
|
|
def add_ndvi(image):
|
|
|
ndvi = image.normalizedDifference(['B8', 'B4']).rename('NDVI')
|
|
|
return image.addBands(ndvi)
|
|
|
|
|
|
def train_classifier():
|
|
|
global gee_classifier
|
|
|
if gee_classifier:
|
|
|
print("✅ Классификатор уже обучен.")
|
|
|
return
|
|
|
|
|
|
print("⏳ Обучение классификатора GEE... Это может занять около минуты.")
|
|
|
try:
|
|
|
desert = ee.FeatureCollection('projects/gen-lang-client-0605302377/assets/kalmykia_desert_samples').map(lambda f: f.set('class', 0))
|
|
|
solonchak = ee.FeatureCollection('projects/gen-lang-client-0605302377/assets/kalmykia_solonchak_samples').map(lambda f: f.set('class', 1))
|
|
|
arid = ee.FeatureCollection('projects/gen-lang-client-0605302377/assets/kalmykia_arid_samples').map(lambda f: f.set('class', 2))
|
|
|
greenery = ee.FeatureCollection('projects/gen-lang-client-0605302377/assets/kalmykia_greenery_samples').map(lambda f: f.set('class', 3))
|
|
|
water = ee.FeatureCollection('projects/gen-lang-client-0605302377/assets/kalmykia_water_samples').map(lambda f: f.set('class', 4))
|
|
|
|
|
|
training_points = desert.merge(solonchak).merge(arid).merge(greenery).merge(water)
|
|
|
roi_for_training = training_points.geometry().bounds()
|
|
|
|
|
|
image_for_training = ee.ImageCollection('COPERNICUS/S2_SR_HARMONIZED') \
|
|
|
.filterDate('2023-06-01', '2023-09-01') \
|
|
|
.filterBounds(roi_for_training) \
|
|
|
.filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 10)) \
|
|
|
.map(add_ndvi) \
|
|
|
.median() \
|
|
|
.clip(roi_for_training)
|
|
|
|
|
|
training_data = image_for_training.select(bands_for_training).sampleRegions(
|
|
|
collection=training_points, properties=['class'], scale=10, tileScale=4)
|
|
|
|
|
|
gee_classifier = ee.Classifier.smileRandomForest(numberOfTrees=50).train(
|
|
|
features=training_data, classProperty='class', inputProperties=bands_for_training)
|
|
|
print("✅ Модель-классификатор успешно обучена на 5 классах!")
|
|
|
except Exception as e:
|
|
|
print(f"🔴 Критическая ошибка во время обучения модели: {e}")
|
|
|
|
|
|
train_classifier()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def get_region_info(region_name):
|
|
|
regions = {
|
|
|
"Элиста (тестовая область)": {"center": [46.307, 44.258], "bbox": [44.15, 46.25, 44.35, 46.35]},
|
|
|
"Озеро Сарпа (Калмыкия)": {"center": [47.85, 44.75], "bbox": [44.65, 47.80, 44.85, 47.90]},
|
|
|
"Арзгир (Ставроп. край)": {"center": [45.38, 44.22], "bbox": [44.12, 45.33, 44.32, 45.43]},
|
|
|
"Будённовск (Ставроп. край)": {"center": [44.78, 44.15], "bbox": [44.05, 44.73, 44.25, 44.83]},
|
|
|
"Москва (тестовая область)": {"center": [55.75, 37.61], "bbox": [37.5, 55.7, 37.7, 55.8]}
|
|
|
}
|
|
|
return regions.get(region_name)
|
|
|
|
|
|
def generate_classified_map(region_info, year, classifier):
|
|
|
if not classifier: return None, "Ошибка: Классификатор не обучен."
|
|
|
print(f"-- Запрос на генерацию карты для {year} года...")
|
|
|
roi_to_classify = ee.Geometry.Rectangle(region_info['bbox'])
|
|
|
collection = ee.ImageCollection('COPERNICUS/S2_SR_HARMONIZED') \
|
|
|
.filterDate(f'{year}-06-01', f'{year}-09-01') \
|
|
|
.filterBounds(roi_to_classify) \
|
|
|
.filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 25))
|
|
|
if collection.size().getInfo() == 0: return None, f"⚠️ Не найдено снимков за {year} год."
|
|
|
image_to_classify = collection.map(add_ndvi).median().clip(roi_to_classify)
|
|
|
classified_image = image_to_classify.classify(classifier)
|
|
|
class_palette = ['#e3a25a', '#ffffff', '#ffff00', '#00ff00', '#0000FF']
|
|
|
vis_params = {'min': 0, 'max': 4, 'palette': class_palette}
|
|
|
map_id = classified_image.getMapId(vis_params)
|
|
|
print(f"-- ✅ Карта для {year} года сгенерирована.")
|
|
|
return {"center": region_info['center'], "tile_url": map_id['tile_fetcher'].url_format, "year": year}, None
|
|
|
|
|
|
def create_map_iframe_html(map_data, region_name):
|
|
|
center, tile_url, year = map_data['center'], map_data['tile_url'], map_data['year']
|
|
|
iframe_content = f"""
|
|
|
<!DOCTYPE html><html><head><title>Карта {year}</title><meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1.0"><link rel="stylesheet" href="https://unpkg.com/leaflet@1.7.1/dist/leaflet.css" /><script src="https://unpkg.com/leaflet@1.7.1/dist/leaflet.js"></script><style>html, body, #map {{ height: 100%; width: 100%; margin: 0; padding: 0; }}</style></head><body><div id="map"></div><script>
|
|
|
var map = L.map('map').setView([{center[0]}, {center[1]}], 12);
|
|
|
L.tileLayer('https://{{s}}.tile.openstreetmap.org/{{z}}/{{x}}/{{y}}.png', {{ attribution: '© OpenStreetMap' }}).addTo(map);
|
|
|
L.tileLayer(`{tile_url}`, {{ attribution: 'Google Earth Engine' }}).addTo(map);
|
|
|
</script></body></html>"""
|
|
|
escaped_html = html.escape(iframe_content)
|
|
|
return f'<iframe srcdoc="{escaped_html}" style="width: 100%; height: 500px; border: 1px solid #ccc;"></iframe>'
|
|
|
|
|
|
def process_and_display_maps(region_name, year1, year2, year3):
|
|
|
if not gee_classifier: return "Ошибка: модель не обучена.", "", ""
|
|
|
region_info = get_region_info(region_name)
|
|
|
years = sorted(list(set([year1, year2, year3])))
|
|
|
outputs, messages = [], []
|
|
|
print(f"\n🚀 Новый запрос! Регион: {region_name}, Годы: {years}")
|
|
|
for year in years:
|
|
|
map_data, msg = generate_classified_map(region_info, year, gee_classifier)
|
|
|
if map_data: outputs.append(create_map_iframe_html(map_data, region_name))
|
|
|
else: outputs.append(f"<p style='text-align:center; padding-top: 200px;'>{msg}</p>")
|
|
|
if msg: messages.append(msg)
|
|
|
while len(outputs) < 3: outputs.append(None)
|
|
|
final_message = "✅ Готово. " + " ".join(messages)
|
|
|
return outputs[0], outputs[1], outputs[2], final_message
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
with gr.Blocks(css=".gradio-container {max-width: 1200px !important;}") as demo:
|
|
|
gr.Markdown("# 🛰️ Анализ почвенного покрова")
|
|
|
gr.Markdown("Выберите регион и три года для анализа. Карты появятся ниже.")
|
|
|
with gr.Row():
|
|
|
with gr.Column(scale=1):
|
|
|
region_dropdown = gr.Dropdown(
|
|
|
label="Регион",
|
|
|
choices=["Элиста (тестовая область)", "Озеро Сарпа (Калмыкия)", "Арзгир (Ставроп. край)", "Будённовск (Ставроп. край)", "Москва (тестовая область)"],
|
|
|
value="Элиста (тестовая область)")
|
|
|
year1_slider = gr.Slider(label="Год 1", minimum=2019, maximum=2025, step=1, value=2019)
|
|
|
year2_slider = gr.Slider(label="Год 2", minimum=2019, maximum=2025, step=1, value=2021)
|
|
|
year3_slider = gr.Slider(label="Год 3", minimum=2019, maximum=2025, step=1, value=2023)
|
|
|
submit_button = gr.Button("Сгенерировать карты", variant="primary")
|
|
|
status_message = gr.Markdown()
|
|
|
with gr.Row():
|
|
|
map1_output = gr.HTML()
|
|
|
map2_output = gr.HTML()
|
|
|
map3_output = gr.HTML()
|
|
|
submit_button.click(
|
|
|
fn=process_and_display_maps,
|
|
|
inputs=[region_dropdown, year1_slider, year2_slider, year3_slider],
|
|
|
outputs=[map1_output, map2_output, map3_output, status_message])
|
|
|
|
|
|
|
|
|
print("\n--- Запуск Gradio интерфейса ---")
|
|
|
demo.launch() |
