init

.gitattributes

@@ -0,0 +1 @@
+*.pkl filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -0,0 +1,13 @@
+FROM python:3.12-slim
+
+WORKDIR /app
+
+COPY requirements.txt .
+
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY . .
+
+EXPOSE 8000
+
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]

Makefile

@@ -0,0 +1,9 @@
+run:
+	docker run -p 8001:8000 --name lsp-container lsp-api
+
+clean:
+	docker stop lsp-container
+	docker rm lsp-container
+
+build:
+	docker build -t lsp-api .

main.py

@@ -0,0 +1,78 @@
+import pickle
+import numpy as np
+import io
+import math
+from fastapi import FastAPI, File, UploadFile, HTTPException
+from PIL import Image
+
+app = FastAPI(
+    title="Peruvian Sign Language (LSP) Recognition API",
+    description="Sube una imagen de una seña del alfabeto de la LSP para predecir la letra correspondiente usando un Mapa Autoorganizado (SOM).",
+    version="1.0.0"
+)
+
+try:
+    with open('lsp_som_model.pkl', 'rb') as f:
+        model_data = pickle.load(f)
+    som = model_data['som']
+    label_map = model_data['label_map']
+    CLASSES = model_data['classes'] # La lista ['A', 'B', 'C', ...]
+    IMG_SIZE = model_data['img_size'] # El tamaño de la imagen
+    print("✅ Modelo y activos cargados exitosamente.")
+    print(f"    - Clases reconocidas: {CLASSES}")
+    print(f"    - Tamaño de imagen esperado: {IMG_SIZE}x{IMG_SIZE}")
+except FileNotFoundError:
+    print("❌ ERROR: No se encontró el archivo del modelo 'lsp_som_model.pkl'.")
+    som = None
+
+def preprocess_image_from_bytes(image_bytes: bytes):
+    try:
+        img = Image.open(io.BytesIO(image_bytes)).convert('L') # 'L' para escala de grises
+        img = img.resize((IMG_SIZE, IMG_SIZE))
+        img_array = np.array(img)
+        img_normalized = img_array / 255.0
+        return img_normalized.flatten()
+    except Exception as e:
+        raise HTTPException(status_code=400, detail=f"Archivo de imagen inválido. Error: {e}")
+
+@app.get("/", tags=["Status"])
+def read_root():
+    return {"status": "ok", "message": "API de Reconocimiento de LSP!!"}
+
+@app.post("/predict", tags=["Prediction"])
+async def predict_sign(file: UploadFile = File(..., description="Un archivo de imagen de una seña de la LSP.")):
+    if not som:
+        raise HTTPException(status_code=503, detail="El modelo no está cargado.")
+
+    image_bytes = await file.read()
+
+    feature_vector = preprocess_image_from_bytes(image_bytes)
+
+    winner_neuron = som.winner(feature_vector)
+
+    predicted_index = label_map.get(winner_neuron, -1)
+
+    # Vecino mas cercano para prediccion
+    is_best_guess = False
+    if predicted_index == -1:
+        is_best_guess = True
+        min_dist = float('inf')
+        for mapped_pos, mapped_label in label_map.items():
+            dist = math.sqrt((winner_neuron[0] - mapped_pos[0])**2 + (winner_neuron[1] - mapped_pos[1])**2)
+            if dist < min_dist:
+                min_dist = dist
+                predicted_index = mapped_label
+    
+    if predicted_index != -1:
+        predicted_letter = CLASSES[predicted_index]
+        prediction_type = "Nearest Neighbor" if is_best_guess else "Direct Match"
+    else:
+        predicted_letter = "Unknown"
+        prediction_type = "Error (No Mapped Neurons Found)"
+
+    return {
+        "filename": file.filename,
+        "predicted_letter": predicted_letter,
+        "prediction_type": prediction_type,
+        "winner_neuron_on_map": [int(coord) for coord in winner_neuron]
+    }

requirements.txt

@@ -0,0 +1,6 @@
+fastapi
+uvicorn[standard]
+python-multipart
+minisom
+numpy
+Pillow