Model_Generator_VGG_Face_to_ONNX_TRUNCATED.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "machine_shape": "hm",
      "gpuType": "A100"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "**2060.io**\n",
        "\n",
        "Open-source tools that enable Institutions building Verifiable Credentials based services"
      ],
      "metadata": {
        "id": "cGjdxSwTnqt3"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Truncated VGG-Face ONNX File generation script"
      ],
      "metadata": {
        "id": "M61yeixOnRDT"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Download the Original VGG-Face model Weights"
      ],
      "metadata": {
        "id": "4rI7eGTFneJh"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "T20FJBW0ZJ2P",
        "outputId": "af2d4652-fcd1-49ce-83fa-29624b1d7198"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2024-11-25 15:08:32--  https://github.com/serengil/deepface_models/releases/download/v1.0/vgg_face_weights.h5\n",
            "Resolving github.com (github.com)... 140.82.114.4\n",
            "Connecting to github.com (github.com)|140.82.114.4|:443... connected.\n",
            "HTTP request sent, awaiting response... 302 Found\n",
            "Location: https://objects.githubusercontent.com/github-production-release-asset-2e65be/382368840/6bf9ba80-db5a-11eb-8b48-84b0d02a64ec?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=releaseassetproduction%2F20241125%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20241125T150833Z&X-Amz-Expires=300&X-Amz-Signature=c85578a324cb0c0fa9f04c599a0b44ad4191696dcec153aaeb2e7b9bcbc77e85&X-Amz-SignedHeaders=host&response-content-disposition=attachment%3B%20filename%3Dvgg_face_weights.h5&response-content-type=application%2Foctet-stream [following]\n",
            "--2024-11-25 15:08:33--  https://objects.githubusercontent.com/github-production-release-asset-2e65be/382368840/6bf9ba80-db5a-11eb-8b48-84b0d02a64ec?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=releaseassetproduction%2F20241125%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20241125T150833Z&X-Amz-Expires=300&X-Amz-Signature=c85578a324cb0c0fa9f04c599a0b44ad4191696dcec153aaeb2e7b9bcbc77e85&X-Amz-SignedHeaders=host&response-content-disposition=attachment%3B%20filename%3Dvgg_face_weights.h5&response-content-type=application%2Foctet-stream\n",
            "Resolving objects.githubusercontent.com (objects.githubusercontent.com)... 185.199.108.133, 185.199.109.133, 185.199.110.133, ...\n",
            "Connecting to objects.githubusercontent.com (objects.githubusercontent.com)|185.199.108.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 580085408 (553M) [application/octet-stream]\n",
            "Saving to: ‘vgg_face_weights.h5’\n",
            "\n",
            "vgg_face_weights.h5 100%[===================>] 553.21M   105MB/s    in 5.2s    \n",
            "\n",
            "2024-11-25 15:08:38 (107 MB/s) - ‘vgg_face_weights.h5’ saved [580085408/580085408]\n",
            "\n"
          ]
        }
      ],
      "source": [
        "!wget https://github.com/serengil/deepface_models/releases/download/v1.0/vgg_face_weights.h5"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Load the dependencies"
      ],
      "metadata": {
        "id": "Ezj7PjyXoMKt"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "tf_version = int(tf.__version__.split(\".\", maxsplit=1)[0])\n",
        "\n",
        "if tf_version == 1:\n",
        "    from keras.models import Model, Sequential\n",
        "    from keras.layers import (\n",
        "        Convolution2D,\n",
        "        ZeroPadding2D,\n",
        "        MaxPooling2D,\n",
        "        Flatten,\n",
        "        Dropout,\n",
        "        Activation,\n",
        "    )\n",
        "else:\n",
        "    from tensorflow.keras.models import Model, Sequential\n",
        "    from tensorflow.keras.layers import (\n",
        "        Convolution2D,\n",
        "        ZeroPadding2D,\n",
        "        MaxPooling2D,\n",
        "        Flatten,\n",
        "        Dropout,\n",
        "        Activation,\n",
        "    )"
      ],
      "metadata": {
        "id": "Vd1MsjkF3sM2"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Define the architecture of the model (.h5 file does not include the architecture)\n",
        "\n",
        "**Note**: The VGG-Face has a last layer with a SoftMax, so we have to remove it."
      ],
      "metadata": {
        "id": "lwoS0tdVoWYe"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import ZeroPadding2D, Convolution2D, MaxPooling2D, Dropout, Flatten, Activation\n",
        "\n",
        "def base_model() -> Sequential:\n",
        "    \"\"\"\n",
        "    Base model of VGG-Face being used for classification - not to find embeddings\n",
        "    Returns:\n",
        "        model (Sequential): model was trained to classify 2622 identities\n",
        "    \"\"\"\n",
        "    model = Sequential()\n",
        "    model.add(ZeroPadding2D((1, 1), input_shape=(224, 224, 3)))\n",
        "    model.add(Convolution2D(64, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(64, (3, 3), activation=\"relu\"))\n",
        "    model.add(MaxPooling2D((2, 2), strides=(2, 2)))\n",
        "\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(128, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(128, (3, 3), activation=\"relu\"))\n",
        "    model.add(MaxPooling2D((2, 2), strides=(2, 2)))\n",
        "\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(256, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(256, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(256, (3, 3), activation=\"relu\"))\n",
        "    model.add(MaxPooling2D((2, 2), strides=(2, 2)))\n",
        "\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(512, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(512, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(512, (3, 3), activation=\"relu\"))\n",
        "    model.add(MaxPooling2D((2, 2), strides=(2, 2)))\n",
        "\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(512, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(512, (3, 3), activation=\"relu\"))\n",
        "    model.add(ZeroPadding2D((1, 1)))\n",
        "    model.add(Convolution2D(512, (3, 3), activation=\"relu\"))\n",
        "    model.add(MaxPooling2D((2, 2), strides=(2, 2)))\n",
        "\n",
        "    model.add(Convolution2D(4096, (7, 7), activation=\"relu\"))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Convolution2D(4096, (1, 1), activation=\"relu\"))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Convolution2D(2622, (1, 1)))\n",
        "    model.add(Flatten())\n",
        "    model.add(Activation(\"softmax\"))\n",
        "\n",
        "    return model\n",
        "\n",
        "# Create the model and load the weights\n",
        "model = base_model()\n",
        "model.load_weights('vgg_face_weights.h5')\n",
        "\n",
        "vgg_face_descriptor = Model(inputs=model.layers[0].input, outputs=model.layers[-2].output)\n",
        "\n",
        "# Save the model in TensorFlow SavedModel format\n",
        "#model.save('saved_vgg_face_model')\n",
        "tf.saved_model.save(model, 'saved_vgg_face_model')\n",
        "tf.saved_model.save(vgg_face_descriptor, 'saved_vgg_face_model_truncated')"
      ],
      "metadata": {
        "id": "Tj7f2dp-jIyJ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b118628f-62d4-46a3-b4bb-a3f51504d5ba"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/keras/src/layers/reshaping/zero_padding2d.py:72: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
            "  super().__init__(**kwargs)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Install Tensorflow to ONNX package"
      ],
      "metadata": {
        "id": "35tN8grho8I1"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install -U tf2onnx"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 620
        },
        "id": "KwVbsa36m8Xh",
        "outputId": "fcd8e1c9-8d10-462b-e112-ff1b0e589951"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting tf2onnx\n",
            "  Downloading tf2onnx-1.16.1-py3-none-any.whl.metadata (1.3 kB)\n",
            "Requirement already satisfied: numpy>=1.14.1 in /usr/local/lib/python3.10/dist-packages (from tf2onnx) (1.26.4)\n",
            "Collecting onnx>=1.4.1 (from tf2onnx)\n",
            "  Downloading onnx-1.17.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (16 kB)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.10/dist-packages (from tf2onnx) (2.32.3)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.10/dist-packages (from tf2onnx) (1.16.0)\n",
            "Requirement already satisfied: flatbuffers>=1.12 in /usr/local/lib/python3.10/dist-packages (from tf2onnx) (24.3.25)\n",
            "Collecting protobuf~=3.20 (from tf2onnx)\n",
            "  Downloading protobuf-3.20.3-cp310-cp310-manylinux_2_12_x86_64.manylinux2010_x86_64.whl.metadata (679 bytes)\n",
            "Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests->tf2onnx) (3.4.0)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests->tf2onnx) (3.10)\n",
            "Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests->tf2onnx) (2.2.3)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests->tf2onnx) (2024.8.30)\n",
            "Downloading tf2onnx-1.16.1-py3-none-any.whl (455 kB)\n",
            "\u001b[2K   \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m455.8/455.8 kB\u001b[0m \u001b[31m15.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hDownloading onnx-1.17.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (16.0 MB)\n",
            "\u001b[2K   \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m16.0/16.0 MB\u001b[0m \u001b[31m108.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hDownloading protobuf-3.20.3-cp310-cp310-manylinux_2_12_x86_64.manylinux2010_x86_64.whl (1.1 MB)\n",
            "\u001b[2K   \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.1/1.1 MB\u001b[0m \u001b[31m59.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hInstalling collected packages: protobuf, onnx, tf2onnx\n",
            "  Attempting uninstall: protobuf\n",
            "    Found existing installation: protobuf 4.25.5\n",
            "    Uninstalling protobuf-4.25.5:\n",
            "      Successfully uninstalled protobuf-4.25.5\n",
            "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
            "grpcio-status 1.62.3 requires protobuf>=4.21.6, but you have protobuf 3.20.3 which is incompatible.\u001b[0m\u001b[31m\n",
            "\u001b[0mSuccessfully installed onnx-1.17.0 protobuf-3.20.3 tf2onnx-1.16.1\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "application/vnd.colab-display-data+json": {
              "pip_warning": {
                "packages": [
                  "google"
                ]
              },
              "id": "b3a38e4d09054f0fa47c83ddd56db1c8"
            }
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Convert the model to ONNX"
      ],
      "metadata": {
        "id": "6e6dvZGzpIfb"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!python -m tf2onnx.convert --saved-model /content/saved_vgg_face_model_truncated --output /content/saved_vgg_face_model_truncated.onnx --opset 13\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Nd1RN21EnMHp",
        "outputId": "1d27eb40-3236-4c99-b3e6-bfb122039fdb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2024-11-25 15:11:30.480801: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
            "2024-11-25 15:11:30.502410: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:8454] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
            "2024-11-25 15:11:30.508993: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1452] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
            "2024-11-25 15:11:31.581233: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n",
            "/usr/lib/python3.10/runpy.py:126: RuntimeWarning: 'tf2onnx.convert' found in sys.modules after import of package 'tf2onnx', but prior to execution of 'tf2onnx.convert'; this may result in unpredictable behaviour\n",
            "  warn(RuntimeWarning(msg))\n",
            "2024-11-25 15:11:33.347687: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:47] Overriding orig_value setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n",
            "2024-11-25 15:11:33,350 - WARNING - '--tag' not specified for saved_model. Using --tag serve\n",
            "2024-11-25 15:11:34,006 - INFO - Signatures found in model: [serving_default].\n",
            "2024-11-25 15:11:34,006 - WARNING - '--signature_def' not specified, using first signature: serving_default\n",
            "2024-11-25 15:11:34,006 - INFO - Output names: ['output_0']\n",
            "WARNING: All log messages before absl::InitializeLog() is called are written to STDERR\n",
            "I0000 00:00:1732547494.012332    1908 devices.cc:67] Number of eligible GPUs (core count >= 8, compute capability >= 0.0): 1\n",
            "I0000 00:00:1732547515.609307    1908 devices.cc:67] Number of eligible GPUs (core count >= 8, compute capability >= 0.0): 1\n",
            "2024-11-25 15:12:10,555 - INFO - Using tensorflow=2.17.1, onnx=1.17.0, tf2onnx=1.16.1/15c810\n",
            "2024-11-25 15:12:10,555 - INFO - Using opset <onnx, 13>\n",
            "2024-11-25 15:12:18,985 - INFO - Computed 0 values for constant folding\n",
            "2024-11-25 15:12:30,282 - INFO - Optimizing ONNX model\n",
            "2024-11-25 15:12:32,529 - INFO - After optimization: Cast -1 (1->0), Const +1 (17->18), Identity -2 (2->0), Reshape +2 (1->3), Transpose -41 (42->1)\n",
            "2024-11-25 15:12:33,400 - INFO - \n",
            "2024-11-25 15:12:33,400 - INFO - Successfully converted TensorFlow model /content/saved_vgg_face_model_truncated to ONNX\n",
            "2024-11-25 15:12:33,400 - INFO - Model inputs: ['inputs']\n",
            "2024-11-25 15:12:33,400 - INFO - Model outputs: ['output_0']\n",
            "2024-11-25 15:12:33,400 - INFO - ONNX model is saved at /content/saved_vgg_face_model_truncated.onnx\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Now the model shoud be in the files**. Just download it.  *saved_vgg_face_model_truncated.onnx*"
      ],
      "metadata": {
        "id": "N8pv4C00pQ7d"
      }
    }
  ]
}