Caatinga ML — API de Segmentação

import base64, requests

# Lê e codifica as imagens
def to_b64(path):
    with open(path, "rb") as f:
        return base64.b64encode(f.read()).decode()

payload = {
    "alert_id":   "SAD_20240115_T24MVB_0042",
    "bbox":       [-40.3821, -9.7654, -40.2190, -9.6023],
    "before_png": to_b64("before.png"),
    "after_png":  to_b64("after.png"),
    "model_name": "resnet34",
}

r = requests.post("http://localhost:8000/predict", json=payload)
r.raise_for_status()
data = r.json()

print(f"Score: {data['prediction_score']:.2%}")
print(f"Modelo usado: {data['model_name']}")
print(f"Componentes: {len(data['debug_components'])} mancha(s)")

# Salva a máscara PNG
import io
from PIL import Image
png = base64.b64decode(data["mask_png_base64"])
Image.open(io.BytesIO(png)).save("mask.png")

# Salva o GeoJSON
import json
with open("mask.geojson", "w") as f:
    json.dump(data["mask_geojson"], f, indent=2)

// Lê arquivos via FileReader e chama o endpoint
async function toBase64(file) {
  return new Promise((resolve, reject) => {
    const r = new FileReader();
    r.onload  = () => resolve(r.result.split(',')[1]);
    r.onerror = reject;
    r.readAsDataURL(file);
  });
}

async function predict(beforeFile, afterFile) {
  const [beforeB64, afterB64] = await Promise.all([
    toBase64(beforeFile),
    toBase64(afterFile),
  ]);

  const res = await fetch('http://localhost:8000/predict', {
    method:  'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      alert_id:   'SAD_20240115_T24MVB_0042',
      bbox:       [-40.3821, -9.7654, -40.2190, -9.6023],
      before_png: beforeB64,
      after_png:  afterB64,
      model_name: 'resnet34',
    }),
  });

  if (!res.ok) throw new Error(await res.text());
  const data = await res.json();

  console.log(`Score: ${(data.prediction_score * 100).toFixed(1)}%`);
  console.log(`Modelo: ${data.model_name}`);
  return data;
}