Deixar apenas conexão via readme

.gitattributes

@@ -1,38 +0,0 @@
-pip install git-filter-repo
-*.7z filter=lfs diff=lfs merge=lfs -text
-*.arrow filter=lfs diff=lfs merge=lfs -text
-*.bin filter=lfs diff=lfs merge=lfs -text
-*.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
-*.ftz filter=lfs diff=lfs merge=lfs -text
-*.gz filter=lfs diff=lfs merge=lfs -text
-*.h5 filter=lfs diff=lfs merge=lfs -text
-*.joblib filter=lfs diff=lfs merge=lfs -text
-*.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.mlmodel filter=lfs diff=lfs merge=lfs -text
-*.model filter=lfs diff=lfs merge=lfs -text
-*.msgpack filter=lfs diff=lfs merge=lfs -text
-*.npy filter=lfs diff=lfs merge=lfs -text
-*.npz filter=lfs diff=lfs merge=lfs -text
-*.onnx filter=lfs diff=lfs merge=lfs -text
-*.ot filter=lfs diff=lfs merge=lfs -text
-*.parquet filter=lfs diff=lfs merge=lfs -text
-*.pb filter=lfs diff=lfs merge=lfs -text
-*.pickle filter=lfs diff=lfs merge=lfs -text
-*.pkl filter=lfs diff=lfs merge=lfs -text
-*.pt filter=lfs diff=lfs merge=lfs -text
-*.pth filter=lfs diff=lfs merge=lfs -text
-*.rar filter=lfs diff=lfs merge=lfs -text
-*.safetensors filter=lfs diff=lfs merge=lfs -text
-saved_model/**/* filter=lfs diff=lfs merge=lfs -text
-*.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
-*.tflite filter=lfs diff=lfs merge=lfs -text
-*.tgz filter=lfs diff=lfs merge=lfs -text
-*.wasm filter=lfs diff=lfs merge=lfs -text
-*.xz filter=lfs diff=lfs merge=lfs -text
-*.zip filter=lfs diff=lfs merge=lfs -text
-*.zst filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text
-models/checkpoints/*.keras filter=lfs diff=lfs merge=lfs -text
-*.keras filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -1,20 +0,0 @@
-FROM python:3.13.3
-
-RUN useradd -m -u 1000 user
-
-RUN apt-get update && apt-get install -y \
-    python3 \
-    python3-pip \
-    libgl1
-
-USER user
-ENV PATH="/home/user/.local/bin:$PATH"
-
-COPY . /app
-
-WORKDIR /app
-
-RUN pip install --no-cache-dir --upgrade -r requirements.txt
-
-COPY --chown=user . /app
-CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -1,84 +0,0 @@
-import uvicorn
-from fastapi import Body, FastAPI, UploadFile, File, Response 
-from fastapi.responses import JSONResponse
-from fastapi.middleware.cors import CORSMiddleware
-import traceback
-import numpy as np 
-import json 
-
-from detect import DengueDetector
-from predict import DenguePredictor  
-
-def default_json_serializer(obj):
-    if isinstance(obj, np.integer):
-        return int(obj)
-    elif isinstance(obj, np.floating):
-        return float(obj)
-    elif isinstance(obj, np.ndarray):
-        return obj.tolist()
-    raise TypeError(f"Object of type {obj.__class__.__name__} is not JSON serializable")
-
-detector: DengueDetector = None
-predictor: DenguePredictor = None
-
-app = FastAPI()
-
-# --- Crie um evento de startup para carregar os modelos ---
-@app.on_event("startup")
-async def startup_event():
-    global detector, predictor
-    print("Executando evento de startup: Carregando os módulos de IA...")
-    detector = DengueDetector()
-    predictor = DenguePredictor()
-    print("Módulos de IA carregados com sucesso. API pronta.")
-
-# --- CORS ---
-origins = ["https://previdengue.vercel.app", "http://localhost:3000", "*"]
-app.add_middleware(
-    CORSMiddleware,
-    allow_origins=origins,
-    allow_credentials=True,
-    allow_methods=["*"],
-    allow_headers=["*"]
-)
-
-# --- Rotas ---
-@app.get("/")
-def health_check():
-    return {"status": "ok", "message": "API de Dengue rodando!"}
-
-@app.post("/detect/")
-async def detect(file: UploadFile = File(...)):
-    if detector is None:
-        return JSONResponse(status_code=503, content={"error": "Detector ainda não foi inicializado."})
-    try:
-        content = await file.read()
-        result = detector.detect_image(content)  
-        return JSONResponse(content=result)
-    except Exception as e:
-        return JSONResponse(status_code=500, content={"error": str(e)})
-
-
-@app.post("/predict/")
-async def predict_dengue_route(payload: dict = Body(...)):
-    if predictor is None:
-        return JSONResponse(status_code=503, content={"error": "Preditor ainda não foi inicializado."})
-    try:
-        ibge_code_str = payload.get("ibge_code")
-        if ibge_code_str is None:
-            raise ValueError("O campo 'ibge_code' é obrigatório.")
-        
-        ibge_code = int(ibge_code_str)
-        result = predictor.predict(ibge_code)
-        
-        json_content = json.dumps(result, default=default_json_serializer)
-        
-        return Response(content=json_content, media_type="application/json")
-
-    except Exception as e:
-        tb_str = traceback.format_exc()
-        print(tb_str) 
-        return JSONResponse(status_code=500, content={
-            "error": str(e),
-            "traceback": tb_str 
-        })

detect.py

@@ -1,54 +0,0 @@
-from collections import Counter
-import numpy as np
-from PIL import Image
-from io import BytesIO
-from ultralytics import YOLO
-
-class DengueDetector:
-    def __init__(self, model_path="./models/DetectsmallTest1.pt"):
-        self.model = YOLO(model_path)
-        self.names = self.model.names
-
-    def calculate_intensity(self, objects):
-        weights = {"piscina": 9, "caixa_agua": 4, "carro": 1}
-        score = sum(weights.get(obj["class"], 0) for obj in objects)
-        return score
-
-    def detect_image(self, image_bytes):
-        # Carregar imagem da memória
-        img = Image.open(BytesIO(image_bytes)).convert("RGB")
-        img_np = np.array(img)  # YOLO aceita np.array diretamente
-        height, width = img_np.shape[:2]
-
-        # Detectar objetos
-        results = self.model(img_np)
-        result = results[0]
-        boxes = result.boxes
-        class_ids = boxes.cls.tolist()
-        confidences = boxes.conf.tolist()
-        class_names = [self.names[int(cls)] for cls in class_ids]
-        counts = Counter(class_names)
-
-        # Construir lista de detecções
-        detections = []
-        for i in range(len(boxes)):
-            x1, y1, x2, y2 = map(float, boxes.xyxy[i])
-            conf = float(confidences[i])
-            cls_id = int(class_ids[i])
-            detections.append({
-                "class": self.names[cls_id],
-                "confidence": round(conf, 4),
-                "box": {
-                    "x1": x1, "y1": y1, "x2": x2, "y2": y2,
-                    "original_width": width, "original_height": height
-                }
-            })
-
-        intensity_score = self.calculate_intensity(detections)
-
-        return {
-            "total": len(class_ids),
-            "contagem": counts,
-            "objetos": detections,
-            "intensity_score": intensity_score
-        }

models/DetectsmallTest1.pt

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3b4b91b6b7d56edae2d369cb6b7306690a9864a1a51d2bf4f1bd6d2877ccf79c
-size 22508451

models/checkpoints/model_checkpoint_best_city.keras

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3b2f243729883b59055234a45b00f16892e60e5b948003ec3c27dc4d022e88a5
-size 2505166

models/checkpoints/test_checkpoint1.keras

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:f3864073b783dad22b1833e95531c6d78df30556af7131d258aa91e7b0fdc5cc
-size 421650

models/scalers/scaler_dyn_global.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:dce234dbd10d12bc0535d23089905f873dc98235550b601e70b702a8665d682f
-size 907

models/scalers/scaler_static_global.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:9699ae64d444fc4453c6524ad3d4d44f7ca09e8bf4f3439c9f19a82f3745e266
-size 743

models/scalers/scaler_target_global.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:981c50a1405169907c95f8e20555949cca362d672f7ee48992e9c31760f85fd8
-size 799

predict.py

@@ -1,204 +0,0 @@
-import os
-import numpy as np
-import pandas as pd
-import joblib
-import warnings
-from pathlib import Path
-from datetime import timedelta
-import tensorflow as tf
-import matplotlib.pyplot as plt
-import base64
-from io import BytesIO
-from huggingface_hub import hf_hub_download 
-
-warnings.filterwarnings('ignore')
-plt.style.use('seaborn-v0_8-darkgrid')
-
-# --- MUDANÇA: Definição da loss customizada necessária para carregar o modelo ---
-def asymmetric_mse(y_true, y_pred):
-    penalty_factor = 5.0
-    error = y_true - y_pred
-    penalty = tf.where(error > 0, penalty_factor, 1.0)
-    loss = tf.square(error) * penalty
-    return tf.reduce_mean(loss)
-
-class DenguePredictor:
-    def __init__(self, project_root=None):
-        self.project_root = Path(project_root) if project_root else Path(__file__).resolve().parent
-        # --- MUDANÇA: Constantes do modelo alinhadas com o treinamento final ---
-        self.sequence_length = 12
-        self.horizon = 8
-        self.year_min_train = 2014
-        self.year_max_train = 2025
-        self.dynamic_features = [
-            "numero_casos", "casos_velocidade", "casos_aceleracao", "casos_mm_4_semanas",
-            "T2M", "T2M_MAX", "T2M_MIN", "PRECTOTCORR", "RH2M", "ALLSKY_SFC_SW_DWN",
-            "week_sin", "week_cos", "year_norm"
-        ]
-        self.static_features = ["latitude", "longitude"]
-        self.feature_names_pt = {
-            "numero_casos": "Nº de Casos de Dengue", "T2M": "Temperatura Média (°C)",
-            "PRECTOTCORR": "Precipitação (mm)"
-        }
-        self.load_assets()
-
-    def load_assets(self):
-        print("INFO: Carregando todos os ativos da IA (modelo, scalers, dados)...")
-        AI_ASSETS_DIR = self.project_root / "models"
-        
-        # --- MUDANÇA: Baixa o arquivo do Hugging Face para o caminho local ---
-        INFERENCE_PATH = hf_hub_download(
-            repo_id='previdengue/predict_inference_data', 
-            filename='inference_data.parquet', 
-            repo_type='dataset',
-            token=os.environ.get('HF_TOKEN') # Autenticação com o token de acesso
-        )
-        
-        SCALER_DIR = AI_ASSETS_DIR / "scalers"
-        MODEL_PATH = AI_ASSETS_DIR / "checkpoints" / "model_checkpoint_best_city.keras"
-
-        # --- MUDANÇA: Carregamento dos scalers GLOBAIS ---
-        self.scaler_dyn = joblib.load(SCALER_DIR / "scaler_dyn_global.pkl")
-        self.scaler_static = joblib.load(SCALER_DIR / "scaler_static_global.pkl")
-        self.scaler_target = joblib.load(SCALER_DIR / "scaler_target_global.pkl")
-
-        # Lê os dados de inferência do arquivo baixado
-        df_master = pd.read_parquet(INFERENCE_PATH)
-        df_master['codigo_ibge'] = df_master['codigo_ibge'].astype(int)
-        df_master['date'] = pd.to_datetime(df_master['ano'].astype(str) + df_master['semana'].astype(str) + '0', format='%Y%W%w', errors='coerce')
-        df_master = df_master.sort_values(by=['codigo_ibge', 'date']).reset_index(drop=True)
-        self.df_master = df_master
-        self.municipios = df_master[['codigo_ibge', 'municipio']].drop_duplicates().sort_values('codigo_ibge')
-
-        # --- MUDANÇA: Carregamento do modelo com a loss customizada ---
-        self.model = tf.keras.models.load_model(MODEL_PATH, custom_objects={'asymmetric_mse': asymmetric_mse})
-        print("INFO: Ativos da IA carregados com sucesso.")
-
-    def plot_to_base64(self, fig):
-        buf = BytesIO()
-        fig.savefig(buf, format='png', bbox_inches='tight', facecolor='#18181b')
-        buf.seek(0)
-        img_str = base64.b64encode(buf.read()).decode('utf-8')
-        plt.close(fig)
-        return img_str
-
-    def predict(self, ibge_code: int):
-        df_mun = self.df_master[self.df_master['codigo_ibge'] == ibge_code].copy()
-        if df_mun.empty or len(df_mun) < self.sequence_length:
-            raise ValueError(f"Não há dados ou histórico suficiente para o município {ibge_code}")
-
-        municipio_name = self.municipios[self.municipios['codigo_ibge'] == ibge_code].iloc[0]['municipio']
-
-        # 1. Pega a última sequência de dados históricos completos
-        last_complete_sequence = df_mun.dropna(subset=['numero_casos']).tail(self.sequence_length).copy()
-        if len(last_complete_sequence) < self.sequence_length:
-            raise ValueError(f"Histórico insuficiente de casos conhecidos para {ibge_code}")
-
-        # 2. Engenharia de Features na sequência de entrada
-        last_complete_sequence['casos_velocidade'] = last_complete_sequence['numero_casos'].diff().fillna(0)
-        last_complete_sequence['casos_aceleracao'] = last_complete_sequence['casos_velocidade'].diff().fillna(0)
-        last_complete_sequence['casos_mm_4_semanas'] = last_complete_sequence['numero_casos'].rolling(4, min_periods=1).mean()
-        last_complete_sequence['week_sin'] = np.sin(2 * np.pi * last_complete_sequence['semana'] / 52)
-        last_complete_sequence['week_cos'] = np.cos(2 * np.pi * last_complete_sequence['semana'] / 52)
-        last_complete_sequence['year_norm'] = (last_complete_sequence['ano'] - self.year_min_train) / (self.year_max_train - self.year_min_train)
-
-        # 3. Prepara os dados de entrada para o modelo
-        dynamic_input_raw = last_complete_sequence[self.dynamic_features].values
-        static_input_raw = last_complete_sequence[self.static_features].iloc[-1].values.reshape(1, -1)
-        
-        dynamic_input_scaled = self.scaler_dyn.transform(dynamic_input_raw).reshape(1, self.sequence_length, -1)
-        static_input_scaled = self.scaler_static.transform(static_input_raw)
-
-        # 4. Faz a predição (tiro único)
-        predictions_scaled = self.model.predict([dynamic_input_scaled, static_input_scaled], verbose=0)
-        pred_casos_scaled = predictions_scaled[0] # Primeira saída do modelo
-
-        # 5. Inverte a transformação para obter o número de casos reais
-        pred_casos_log = self.scaler_target.inverse_transform(pred_casos_scaled.reshape(1, -1))
-        pred_casos_real = np.expm1(pred_casos_log).flatten()
-        
-        predictions_final = [max(0, round(case)) for case in pred_casos_real]
-
-        # 6. Formata a resposta
-        last_real_date = last_complete_sequence['date'].iloc[-1]
-        predicted_data = [{
-            "date": (last_real_date + timedelta(weeks=i + 1)).strftime('%Y-%m-%d'),
-            "predicted_cases": cases
-        } for i, cases in enumerate(predictions_final)]
-
-        historic_data = [{
-            "date": row['date'].strftime('%Y-%m-%d'),
-            "cases": int(row["numero_casos"]) if pd.notna(row["numero_casos"]) else None
-        } for _, row in df_mun.tail(52).iterrows()]
-
-        # Geração de insights (Análise de Lag)
-        lag_plot_b64, strategic_summary, tipping_points = self.generate_lag_insights(df_mun)
-
-        return {
-            "municipality_name": municipio_name,
-            "historic_data": historic_data,
-            "predicted_data": predicted_data,
-            "insights": {
-                "lag_analysis_plot_base64": lag_plot_b64,
-                "strategic_summary": strategic_summary,
-                "tipping_points": tipping_points
-            }
-        }
-    
-    def generate_lag_insights(self, df_mun):
-        # Renomeia as colunas para a análise e para os gráficos
-        df_analysis = df_mun.rename(columns={"T2M": "Temperatura Média (°C)", "PRECTOTCORR": "Precipitação (mm)"})
-        max_lag = 12
-        cases_col_name = 'numero_casos'
-        lag_features = ['Temperatura Média (°C)', 'Precipitação (mm)']
-        lag_correlations = {}
-
-        # Calcula a correlação para cada feature com diferentes defasagens (lags)
-        for col in lag_features:
-            # Garante que a coluna existe antes de tentar usá-la
-            if col in df_analysis.columns:
-                corrs = [df_analysis[cases_col_name].corr(df_analysis[col].shift(lag)) for lag in range(1, max_lag + 1)]
-                lag_correlations[col] = corrs
-        
-        # Cria a figura para o gráfico
-        fig, ax = plt.subplots(figsize=(10, 6), facecolor='#18181b')
-        ax.set_facecolor('#18181b')
-        
-        # Plota as correlações
-        for feature_name, corrs in lag_correlations.items():
-            ax.plot(range(1, max_lag + 1), corrs, marker='o', linestyle='-', label=feature_name)
-        
-        # Estiliza o gráfico
-        ax.set_title('Análise de Defasagem (Lag)', color='white')
-        ax.set_xlabel('Defasagem (Semanas)', color='white')
-        ax.set_ylabel('Correlação com Casos', color='white')
-        ax.tick_params(colors='white')
-        ax.legend(facecolor='#27272a', edgecolor='gray', labelcolor='white')
-        ax.grid(True, which='both', linestyle='--', linewidth=0.5, color='#444')
-        
-        # Converte o gráfico para base64 para enviar na resposta da API
-        lag_plot_b64 = self.plot_to_base64(fig)
-
-        # Encontra o pico de correlação para cada feature
-        lag_peaks = {
-            feature: (np.argmax(np.abs(corrs)) + 1) if corrs and not all(pd.isna(corrs)) else 'N/A' 
-            for feature, corrs in lag_correlations.items()
-        }
-        temp_lag = lag_peaks.get('Temperatura Média (°C)', 'N/A')
-        rain_lag = lag_peaks.get('Precipitação (mm)', 'N/A')
-
-        # Cria um resumo estratégico
-        summary = (
-            f"A IA identifica a **Temperatura** e a **Precipitação** como os principais gatilhos climáticos. "
-            f"O impacto da temperatura tende a ser máximo após **{temp_lag} semanas**, enquanto o da chuva ocorre após **{rain_lag} semanas**. "
-            "Ações preventivas devem ser intensificadas nesta janela após eventos climáticos extremos."
-        )
-        
-        # Cria os pontos-chave (tipping points)
-        tipping_points = [
-            {"factor": "Temperatura", "value": f"Impacto máximo em {temp_lag} semanas"},
-            {"factor": "Precipitação", "value": f"Impacto máximo em {rain_lag} semanas"},
-            {"factor": "Umidade", "value": "Aumenta a sobrevida do mosquito adulto"},
-        ]
-        
-        return lag_plot_b64, summary, tipping_points

requirements.txt

@@ -1,16 +0,0 @@
-fastapi==0.115.12
-uvicorn==0.34.0
-pillow==11.1.0
-opencv-python-headless
-numpy==2.1.1
-ultralytics==8.3.105
-torch==2.6.0
-python-multipart
-joblib==1.5.1
-pandas==2.2.3
-matplotlib==3.10.3
-tensorflow==2.20.0
-epiweeks==2.3.0
-scikit-learn==1.6.1
-fastparquet
-huggingface_hub