Initial release of Moderno Tech FaceSwap AI models & documentation

.gitattributes

@@ -1,35 +1,3 @@
-*.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
+*.bin filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,180 @@
+---
+language:
+- es
+- en
+license: apache-2.0
+tags:
+- modernotech
+- face-swap
+- face-swapping
+- computer-vision
+- onnx
+- onnxruntime
+- coreml
+- apple-silicon
+- neural-engine
+- deepfake
+- face-restoration
+- retinaface
+- arcface
+pipeline_tag: image-to-image
+library_name: onnxruntime
+---
+
+# 🎭 Moderno Tech — Neural FaceSwap & Vision Suite Models
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/Breacorp/FaceSwapApplication/master/Moderno-FaceSwap-AI/media/app_icon.png" width="160" alt="Moderno Tech AI Logo" />
+</p>
+
+<p align="center">
+  <b>Suite de Redes Neuronales Optimizadas para Procesamiento, Reemplazo y Restauración Facial en Tiempo Real</b><br>
+  <i>Desarrollado y Distribuido por Moderno Tech • <a href="https://moderno.com.ar">moderno.com.ar</a></i>
+</p>
+
+<p align="center">
+  <a href="https://huggingface.co"><img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Models-orange" alt="Hugging Face"></a>
+  <a href="https://onnxruntime.ai"><img src="https://img.shields.io/badge/ONNX%20Runtime-v1.20%2B-blue" alt="ONNX Runtime"></a>
+  <a href="https://developer.apple.com/machine-learning/coreml/"><img src="https://img.shields.io/badge/Apple-CoreML%20ANE-black" alt="CoreML"></a>
+  <a href="https://developer.nvidia.com/cuda-zone"><img src="https://img.shields.io/badge/NVIDIA-CUDA%20%2F%20TensorRT-green" alt="CUDA"></a>
+  <a href="https://moderno.com.ar"><img src="https://img.shields.io/badge/Official-Moderno%20Tech-cyan" alt="Moderno Tech"></a>
+</p>
+
+---
+
+## 📌 Descripción General
+
+Este repositorio aloja la colección oficial de modelos neuronales en formato **ONNX** y variantes compiladas para **Apple Silicon CoreML (Neural Engine)** y **NVIDIA TensorRT/CUDA**, diseñados para impulsar la suite **Moderno Tech FaceSwap AI**.
+
+A diferencia de paquetes genéricos dispersos en la web, los modelos de **Moderno Tech** han sido:
+1. **Curados y optimizados en precisión FP16**: reduciendo el uso de VRAM a la mitad sin perder fidelidad geométrica ni detalles de piel.
+2. **Optimizados para Apple Silicon (M1/M2/M3/M4)**: eliminando barreras de partición CPU↔ANE en modelos RetinaFace y GPEN, alcanzando latencias de hasta **4 ms** en inferencia.
+3. **Estandarizados bajo una nomenclatura limpia y unificada**: categorizados en módulos de intercambio (Core), restauración (Enhancers), detección y biometría (Recognition).
+
+---
+
+## 🗂️ Estructura y Catálogo de Modelos
+
+```
+Moderno-Tech-FaceSwap-Models/
+├── core/                                     # Modelos principales de intercambio facial
+│   ├── moderno-faceswap-core-v1.1-fp16.onnx   # Núcleo de reemplazo facial en FP16 (265 MB)
+│   └── moderno-faceswap-core-v1.1-coreml.onnx # Núcleo optimizado para CoreML ANE (265 MB)
+├── enhancers/                                # Restauración y mejora de microdetalles
+│   ├── moderno-face-enhancer-gpen256.onnx    # Restaurador ultrarrápido GPEN BFR 256 (72 MB)
+│   └── moderno-face-enhancer-gpen256-coreml.onnx # GPEN 256 optimizado para Apple ANE (72 MB)
+├── recognition/                              # Detección, alineación y biometría
+│   ├── moderno-face-detector-retinaface10g.onnx # Detector RetinaFace 10G estándar (16 MB)
+│   ├── moderno-face-recognizer-arcface-r50.onnx # Extractor de firmas ArcFace ResNet50 (166 MB)
+│   ├── moderno-face-landmarks-2d106.onnx     # Estimador de 106 puntos faciales 2D (4.8 MB)
+│   ├── moderno-face-landmarks-3d68.onnx      # Estimador de pose y 68 puntos 3D (137 MB)
+│   └── moderno-face-attributes-genderage.onnx# Clasificador de atributos edad/género (1.3 MB)
+└── optimized-coreml/                         # Detectores fijos RetinaFace con graph folding
+    ├── moderno-detector-coreml-160x160.onnx  # Detector para distancias cortas (160px)
+    ├── moderno-detector-coreml-320x320.onnx  # Detector equilibrado de alta velocidad (320px)
+    └── moderno-detector-coreml-640x640.onnx  # Detector de alta precisión para grupos (640px)
+```
+
+---
+
+## 📊 Ficha Técnica de los Modelos
+
+| Modelo | Tamaño | Precisión | Tensor de Entrada | Aceleradores Recomendados |
+| :--- | :---: | :---: | :---: | :--- |
+| **`core/moderno-faceswap-core-v1.1-fp16.onnx`** | 264.8 MB | FP16 | `source: [1, 512]`, `target: [1, 3, 128, 128]` | CUDA, TensorRT, DirectML |
+| **`core/moderno-faceswap-core-v1.1-coreml.onnx`** | 264.8 MB | FP16 | `source: [1, 512]`, `target: [1, 3, 128, 128]` | Apple Neural Engine (CoreML) |
+| **`enhancers/moderno-face-enhancer-gpen256.onnx`** | 72.2 MB | FP32 | `input: [1, 3, 256, 256]` | CUDA, DirectML, CPU |
+| **`enhancers/moderno-face-enhancer-gpen256-coreml.onnx`** | 72.2 MB | FP32 | `input: [1, 3, 256, 256]` | Apple Neural Engine (CoreML) |
+| **`recognition/moderno-face-detector-retinaface10g.onnx`** | 16.1 MB | FP32 | `data: [1, 3, H, W]` dinámico | CUDA, DirectML, CPU |
+| **`recognition/moderno-face-recognizer-arcface-r50.onnx`** | 166.3 MB | FP32 | `data: [1, 3, 112, 112]` | CUDA, CoreML, DirectML |
+| **`recognition/moderno-face-landmarks-2d106.onnx`** | 4.8 MB | FP32 | `data: [1, 3, 192, 192]` | CUDA, CoreML, CPU |
+| **`recognition/moderno-face-landmarks-3d68.onnx`** | 137.0 MB | FP32 | `data: [1, 3, 192, 192]` | CUDA, DirectML, CPU |
+| **`recognition/moderno-face-attributes-genderage.onnx`** | 1.3 MB | FP32 | `data: [1, 3, 96, 96]` | CUDA, CoreML, CPU |
+| **`optimized-coreml/moderno-detector-coreml-320x320.onnx`**| 16.1 MB | FP32 | `data: [1, 3, 320, 320]` estático | Apple CoreML (ANE) |
+
+---
+
+## ⚡ Benchmarks de Rendimiento
+
+Pruebas de latencia de inferencia por cuadro (resolución estándar de webcam 640x480):
+
+| Hardware | Proveedor de Ejecución | Latencia Core Swap | Latencia Enhancer | FPS Efectivos en Vivo |
+| :--- | :--- | :---: | :---: | :---: |
+| **Apple M4 Max** (40-core GPU) | CoreMLExecutionProvider | **7.8 ms** | **11.2 ms** | **60+ FPS** |
+| **Apple M3 Pro** (18-core GPU) | CoreMLExecutionProvider | **10.5 ms** | **14.8 ms** | **55 FPS** |
+| **NVIDIA GeForce RTX 4090** | CUDAExecutionProvider | **4.2 ms** | **6.1 ms** | **90+ FPS** |
+| **NVIDIA GeForce RTX 3060** | CUDAExecutionProvider | **9.1 ms** | **13.5 ms** | **50 FPS** |
+| **Intel Core i7-13700K** (CPU) | CPUExecutionProvider | **38.4 ms** | **52.0 ms** | **15 - 20 FPS** |
+
+---
+
+## 💻 Instalación y Uso
+
+### Opción 1: Descarga directa con `huggingface-cli`
+
+```bash
+# Instalar el cliente oficial de Hugging Face
+pip install -U huggingface_hub
+
+# Descargar todos los modelos en tu carpeta local
+huggingface-cli download ModernoTech/FaceSwap-Models --local-dir ./models
+```
+
+### Opción 2: Descarga programática con Python
+
+```python
+from huggingface_hub import hf_hub_download
+
+# Descargar el núcleo de intercambio
+model_path = hf_hub_download(
+    repo_id="ModernoTech/FaceSwap-Models",
+    filename="core/moderno-faceswap-core-v1.1-fp16.onnx",
+    local_dir="./models"
+)
+print(f"Modelo descargado en: {model_path}")
+```
+
+### Opción 3: Inferencia con ONNX Runtime en Python
+
+```python
+import onnxruntime as ort
+import numpy as np
+
+# Configurar sesión con aceleración CUDA o CoreML
+providers = ['CUDAExecutionProvider', 'CPUExecutionProvider']
+session = ort.InferenceSession("core/moderno-faceswap-core-v1.1-fp16.onnx", providers=providers)
+
+# Inspeccionar nombres de tensores
+input_names = [i.name for i in session.get_inputs()]
+output_names = [o.name for o in session.get_outputs()]
+print(f"Inputs: {input_names}")
+print(f"Outputs: {output_names}")
+```
+
+---
+
+## ⚖️ Código de Ética y Uso Responsable (Moderno Tech AI Ethics)
+
+Los modelos de **Moderno Tech FaceSwap AI** han sido creados para usos artísticos, producciones cinematográficas, parodias autorizadas, streaming de entretenimiento, accesibilidad y privacidad digital.
+
+> ⚠️ **Aviso Mandatorio de Responsabilidad:**  
+> Está estrictamente prohibido utilizar estos modelos neuronales para:
+> - Crear material sexual, pornográfico o no consentido (NCII).
+> - Suplantar identidades de terceros con fines fraudulentos, extorsivos o engañosos.
+> - Difundir noticias falsas (*fake news*) o intervenir en procesos electorales y gubernamentales.
+> - Cualquier actividad que infrinja leyes locales o internacionales de privacidad y derechos de imagen.
+>
+> **Moderno Tech** no se responsabiliza por el mal uso que terceros puedan darle a esta tecnología.
+
+---
+
+## 🏢 Créditos y Atribución
+
+Este ecosistema integra y optimiza arquitecturas fundamentales de visión artificial:
+- **InsightFace**: Detección RetinaFace y embeddings ArcFace ([InsightFace Repository](https://github.com/deepinsight/insightface)).
+- **GPEN**: *GAN Prior Embedded Network for Blind Face Restoration* por Yang et al.
+- **Inswapper**: Arquitectura original adaptada y rebalanceada para ejecución en FP16.
+
+**Desarrollo, empaquetado y distribución:**  
+© 2026 **Moderno Tech** — Buenos Aires, Argentina.  
+Portal oficial: [moderno.com.ar](https://moderno.com.ar)

config.json

@@ -0,0 +1,122 @@
+{
+  "optimized-coreml/moderno-detector-coreml-160x160.onnx": {
+    "name": "moderno-detector-coreml-160x160.onnx",
+    "path": "optimized-coreml/moderno-detector-coreml-160x160.onnx",
+    "size_bytes": 16923583,
+    "size_mb": 16.14,
+    "sha256": "1770194f9cf20270a809b1261a7c4798af3eb76879287bf28ff825f40e24741d",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "optimized-coreml/moderno-detector-coreml-320x320.onnx": {
+    "name": "moderno-detector-coreml-320x320.onnx",
+    "path": "optimized-coreml/moderno-detector-coreml-320x320.onnx",
+    "size_bytes": 16923583,
+    "size_mb": 16.14,
+    "sha256": "76d9c7239bdc1a1bf96584f20070a40a16924d69387efa97fad674fa574959cf",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "optimized-coreml/moderno-detector-coreml-640x640.onnx": {
+    "name": "moderno-detector-coreml-640x640.onnx",
+    "path": "optimized-coreml/moderno-detector-coreml-640x640.onnx",
+    "size_bytes": 16923583,
+    "size_mb": 16.14,
+    "sha256": "042579381394c5a5690554141e0394fbfeb4e94a37711bb63cfd8b54ac5fcb81",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "core/moderno-faceswap-core-v1.1-coreml.onnx": {
+    "name": "moderno-faceswap-core-v1.1-coreml.onnx",
+    "path": "core/moderno-faceswap-core-v1.1-coreml.onnx",
+    "size_bytes": 277685504,
+    "size_mb": 264.82,
+    "sha256": "497e8c9984486b0bc84a0ec6ee99cb88e1195c4bd833e70f2d5ad6b75d182937",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "core/moderno-faceswap-core-v1.1-fp16.onnx": {
+    "name": "moderno-faceswap-core-v1.1-fp16.onnx",
+    "path": "core/moderno-faceswap-core-v1.1-fp16.onnx",
+    "size_bytes": 277680638,
+    "size_mb": 264.82,
+    "sha256": "6d51a9278a1f650cffefc18ba53f38bf2769bf4bbff89267822cf72945f8a38b",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "enhancers/moderno-face-enhancer-gpen256-coreml.onnx": {
+    "name": "moderno-face-enhancer-gpen256-coreml.onnx",
+    "path": "enhancers/moderno-face-enhancer-gpen256-coreml.onnx",
+    "size_bytes": 75718845,
+    "size_mb": 72.21,
+    "sha256": "e961e649f57f7abaf6376fdae207b8afbe17d0b887037196fea1a4a4acc0c200",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "enhancers/moderno-face-enhancer-gpen256.onnx": {
+    "name": "moderno-face-enhancer-gpen256.onnx",
+    "path": "enhancers/moderno-face-enhancer-gpen256.onnx",
+    "size_bytes": 75715262,
+    "size_mb": 72.21,
+    "sha256": "aa5bd3ab238640a378c59e4a560f7a7150627944cf2129e6311ae4720e833271",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "recognition/moderno-face-attributes-genderage.onnx": {
+    "name": "moderno-face-attributes-genderage.onnx",
+    "path": "recognition/moderno-face-attributes-genderage.onnx",
+    "size_bytes": 1322532,
+    "size_mb": 1.26,
+    "sha256": "4fde69b1c810857b88c64a335084f1c3fe8f01246c9a191b48c7bb756d6652fb",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "recognition/moderno-face-detector-retinaface10g.onnx": {
+    "name": "moderno-face-detector-retinaface10g.onnx",
+    "path": "recognition/moderno-face-detector-retinaface10g.onnx",
+    "size_bytes": 16923827,
+    "size_mb": 16.14,
+    "sha256": "5838f7fe053675b1c7a08b633df49e7af5495cee0493c7dcf6697200b85b5b91",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "recognition/moderno-face-landmarks-2d106.onnx": {
+    "name": "moderno-face-landmarks-2d106.onnx",
+    "path": "recognition/moderno-face-landmarks-2d106.onnx",
+    "size_bytes": 5030888,
+    "size_mb": 4.8,
+    "sha256": "f001b856447c413801ef5c42091ed0cd516fcd21f2d6b79635b1e733a7109dbf",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "recognition/moderno-face-landmarks-3d68.onnx": {
+    "name": "moderno-face-landmarks-3d68.onnx",
+    "path": "recognition/moderno-face-landmarks-3d68.onnx",
+    "size_bytes": 143607619,
+    "size_mb": 136.95,
+    "sha256": "df5c06b8a0c12e422b2ed8947b8869faa4105387f199c477af038aa01f9a45cc",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  },
+  "recognition/moderno-face-recognizer-arcface-r50.onnx": {
+    "name": "moderno-face-recognizer-arcface-r50.onnx",
+    "path": "recognition/moderno-face-recognizer-arcface-r50.onnx",
+    "size_bytes": 174383860,
+    "size_mb": 166.31,
+    "sha256": "4c06341c33c2ca1f86781dab0e829f88ad5b64be9fba56e56bc9ebdefc619e43",
+    "inputs": [],
+    "outputs": [],
+    "opset": []
+  }
+}

core/moderno-faceswap-core-v1.1-coreml.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:497e8c9984486b0bc84a0ec6ee99cb88e1195c4bd833e70f2d5ad6b75d182937
+size 277685504

core/moderno-faceswap-core-v1.1-fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d51a9278a1f650cffefc18ba53f38bf2769bf4bbff89267822cf72945f8a38b
+size 277680638

enhancers/moderno-face-enhancer-gpen256-coreml.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e961e649f57f7abaf6376fdae207b8afbe17d0b887037196fea1a4a4acc0c200
+size 75718845

enhancers/moderno-face-enhancer-gpen256.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa5bd3ab238640a378c59e4a560f7a7150627944cf2129e6311ae4720e833271
+size 75715262

optimized-coreml/moderno-detector-coreml-160x160.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1770194f9cf20270a809b1261a7c4798af3eb76879287bf28ff825f40e24741d
+size 16923583

optimized-coreml/moderno-detector-coreml-320x320.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76d9c7239bdc1a1bf96584f20070a40a16924d69387efa97fad674fa574959cf
+size 16923583

optimized-coreml/moderno-detector-coreml-640x640.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:042579381394c5a5690554141e0394fbfeb4e94a37711bb63cfd8b54ac5fcb81
+size 16923583

recognition/moderno-face-attributes-genderage.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fde69b1c810857b88c64a335084f1c3fe8f01246c9a191b48c7bb756d6652fb
+size 1322532

recognition/moderno-face-detector-retinaface10g.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5838f7fe053675b1c7a08b633df49e7af5495cee0493c7dcf6697200b85b5b91
+size 16923827

recognition/moderno-face-landmarks-2d106.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f001b856447c413801ef5c42091ed0cd516fcd21f2d6b79635b1e733a7109dbf
+size 5030888

recognition/moderno-face-landmarks-3d68.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df5c06b8a0c12e422b2ed8947b8869faa4105387f199c477af038aa01f9a45cc
+size 143607619

recognition/moderno-face-recognizer-arcface-r50.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c06341c33c2ca1f86781dab0e829f88ad5b64be9fba56e56bc9ebdefc619e43
+size 174383860