visualizatizing_featuremaps.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "KSmc9kXd4YWi"
      },
      "outputs": [],
      "source": [
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.models import load_model"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "saved_model = load_model('/content/drive/MyDrive/models/cnn_adagrad_e45_bc16/model.h5')"
      ],
      "metadata": {
        "id": "Njnd-KXS4q0M"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model = saved_model"
      ],
      "metadata": {
        "id": "DuRFeTC85CWD"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BrbIJZg15Jre",
        "outputId": "803d86b1-5a7c-4066-deac-ec1e52f90d8e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv3d (Conv3D)             (None, 20, 98, 98, 64)    1792      \n",
            "                                                                 \n",
            " max_pooling3d (MaxPooling3D  (None, 10, 49, 49, 64)   0         \n",
            " )                                                               \n",
            "                                                                 \n",
            " conv3d_1 (Conv3D)           (None, 8, 47, 47, 128)    221312    \n",
            "                                                                 \n",
            " max_pooling3d_1 (MaxPooling  (None, 4, 23, 23, 128)   0         \n",
            " 3D)                                                             \n",
            "                                                                 \n",
            " conv3d_2 (Conv3D)           (None, 3, 22, 22, 256)    262400    \n",
            "                                                                 \n",
            " max_pooling3d_2 (MaxPooling  (None, 1, 11, 11, 256)   0         \n",
            " 3D)                                                             \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 30976)             0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 4096)              126881792 \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 4096)              0         \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 2048)              8390656   \n",
            "                                                                 \n",
            " dropout_1 (Dropout)         (None, 2048)              0         \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 10)                20490     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 135,778,442\n",
            "Trainable params: 135,778,442\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "get list of all layers\n"
      ],
      "metadata": {
        "id": "JccN1kbA5jaI"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "layer = model.layers\n",
        "layer"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2W79owwo5L5O",
        "outputId": "9369f833-b953-4839-ac84-c9bc2176be5c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<keras.layers.convolutional.conv3d.Conv3D at 0x7fc95feec590>,\n",
              " <keras.layers.pooling.max_pooling3d.MaxPooling3D at 0x7fc9603c3c10>,\n",
              " <keras.layers.convolutional.conv3d.Conv3D at 0x7fc9502aaf90>,\n",
              " <keras.layers.pooling.max_pooling3d.MaxPooling3D at 0x7fc9502ccfd0>,\n",
              " <keras.layers.convolutional.conv3d.Conv3D at 0x7fc9502aa310>,\n",
              " <keras.layers.pooling.max_pooling3d.MaxPooling3D at 0x7fc950279450>,\n",
              " <keras.layers.reshaping.flatten.Flatten at 0x7fc95027e410>,\n",
              " <keras.layers.core.dense.Dense at 0x7fc95027f690>,\n",
              " <keras.layers.regularization.dropout.Dropout at 0x7fc950279ed0>,\n",
              " <keras.layers.core.dense.Dense at 0x7fc95027e750>,\n",
              " <keras.layers.regularization.dropout.Dropout at 0x7fc9502cc1d0>,\n",
              " <keras.layers.core.dense.Dense at 0x7fc95027f590>]"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "filters, biases = model.layers[0].get_weights()\n",
        "print(layer[0].name, filters.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hCS697XH5fWX",
        "outputId": "2805079d-0762-4f71-9164-8f15cc2d88c6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "conv3d (3, 3, 3, 1, 64)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt"
      ],
      "metadata": {
        "id": "6XRjTIjH6eYS"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "fig1 = plt.figure(figsize=(8,12))\n",
        "columns = 8\n",
        "rows = 8\n",
        "n_filters = columns * rows\n",
        "for i in range(1, n_filters +1):\n",
        "    f = filters[:, :, :,0, i-1]\n",
        "    fig1 =plt.subplot(rows, columns, i)\n",
        "    fig1.set_xticks([])  #Turn off axis\n",
        "    fig1.set_yticks([])\n",
        "    plt.imshow(f[:, :, 0], cmap='gray') #Show only the filters from 0th channel (R)\n",
        "    #ix += 1\n",
        "plt.show() "
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 665
        },
        "id": "niyZoXEX6aBB",
        "outputId": "0624f28d-9214-4863-929a-5037abe7a0e2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 576x864 with 64 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdAAAAKICAYAAADAaKsHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dfZDddX02/vdJdrO7SWCzeZiEJJQFUyFSkWIA0aK0RaSUYYoiooItMrbYqgylFLVTLI4VlYJWytgiQgcoAiJVQGmpDyNoLZ2041TT8mDDJoSQhM3zkoTNw7n/gH+cG9jvXt8ft/yc1+vffV/zPp/dc86Vc2bg0+l2uwUATM6Un/cDAID/P1KgABBQoAAQUKAAEFCgABBQoAAQ6JnMcH9/f3fmzJnxsjlz5sTZqqonn3wyzu7atavGx8c7E83Nnj27u3jx4njPtm3b4mxV1YwZM+Ls2rVra/PmzROecerUqd2enkn96X/G+Ph4nK2qeu1rXxtnR0ZGanR09CX/O/74xz+Os1VVRx11VJxdtWpVozN2Op1W/w1aX19fm3i1eS8YGxurXbt2TXjGgYGB7uDgYLxn165dcbaqamhoKM6Ojo7W9u3bJzzjrFmzugsWLIj3PPzww3G2qqrN7q1bt9aOHTsmPOPg4GCrM27ZsiXOVlUdeOCBcfbF3nMm9S46c+bMOu200+IHcvbZZ8fZqqrLL788zj744ION5hYvXlzf+MY34j3//M//HGerqo4++ug4+653vavRXE9PTy1cuDDeMzIyEmerqpYvXx5nly1b1mhu8eLFddddd8V7Dj744Dhb1fz59nyOPfbYVrubavOmUlX1+te/Ps42fY0NDg62et946KGH4mxV1RlnnBFnL7vsskZzCxYsqGuvvTbe86Y3vSnOVlWde+65cfaGG25oNNf2jF/96lfjbFXV5z//+Tj7Yu85vsIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgMCkbmMZHh6u66+/Pl62dOnSOFvV/maFJsbGxuoHP/hBnP/EJz7Ran+bm04GBgYazU2ZMqXVVVRtb38455xz4uxjjz3WaG716tX1oQ99KN7T7ba6KaxmzZoVZ8fGxhrNTZ06tdXf8dFHH42zVVXXXXddnP3e977XaK7b7dbevXvjPffcc0+crapW73dNn0MzZ86sN77xjfGeDRs2xNmqqne/+91x9umnn24098gjj9QJJ5wQ73nFK14RZ6uqPvrRj8bZ3bt3v+DPfAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAwKTuA33sscfqXe96V7xsxYoVcbaq3Z1wa9eubTT32GOP1Tvf+c54z6pVq+JsVdXRRx8dZ//nf/6n0dwhhxxSt9xyS7zniSeeiLNV1eq+1b6+vkZzCxcurI9//OPxno985CNxtqrd3bVvectbGs3Nmzevzj333HjPnXfeGWerqjZu3Bhn9+zZ02hux44dtXz58nhPp9OJs1VVS5YsibPr1q1rNLdx48a66aab4j3ve9/74mxV1aJFi+LsM88802jukEMOqc985jPxnrbdMX369Dg7ZcoLf870CRQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACnW6323y403mqqtpdePnzc1C325030ZAzvuw543Oc8WXPGZ/zi3rGSRUoAPAsX+ECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQKBnMsOdTqfbZtkrXvGKNvGaNWtWnB0ZGanR0dHORHPTp0/vDg4Oxnvmzp0bZ6uqHn/88Ti7Y8eOGh8fn/CMc+fO7Q4PD8d7/uM//iPOVlUtWbIkzq5fv762bds24RnbPleHhobaxGv+/Plx9sknn6wtW7ZMeMbBwcFumz2jo6NxtqrqgAMOiLNr166tzZs3T3jGgYGB7v777x/v2bBhQ5ytqpo6dWqc3bdvX+3bt2/CM86ZM6d74IEHxnvGx8fjbFXVY489Fmd3795de/bsmfCMvb293f7+/njPzJkz42xV1YIFC+Ls6tWra+PGjc97xkkVaFtXXHFFq/xpp50WZ4899thGc4ODg3XuuefGe973vvfF2aqqD37wg3H2+9//fqO54eHhWr58ebyn05nw9fKirrrqqjj7x3/8x612N/WWt7ylVf7CCy+Ms7/7u7/baG7+/Pl1zTXXxHuuvfbaOFtVdemll8bZs846q9Hc/vvvX+94xzviPVdffXWcrWr3j/YtW7Y0mjvwwAPrvvvui/esXbs2zlZVvfvd746zK1eubDTX399fRx55ZLzn137t1+JsVdWHP/zhOHvCCSe84M98hQsAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAIFJXWe2//7713HHHRcv+9GPfhRnq6pOP/30Vvkmdu7cWStWrIjz//mf/9lq/ze+8Y1W+SY2b95ct99+e5w/88wzW+3/3ve+F2e3b9/eaG7x4sV1wQUXxHs+97nPxdmqanVdXNM7Ybdt21bf+ta34j1Lly6Ns1VVRxxxRKt8EzNnzqzjjz8+zn/+859vtf+HP/xhnH3ve9/baG79+vWtnm8LFy6Ms1XV6v1u2bJljebGxsYaX7f4fPbt2xdnq6oefvjhODsyMvKCP/MJFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAKTug904cKF9fGPfzxedvLJJ8fZqmq1u6mDDjqorr322jh/5ZVXttrf7XbjbNO7+dasWVOXXHJJvOdrX/tanK2q+su//Ms4Oz4+3mhudHS0brjhhnjPhg0b4mxV88f5fJr+HWfNmlWnnnpqvOeAAw6Is1XtXo9NzzgwMFC/8iu/Eu9561vfGmerqnp7e+Ps+vXrG83t3LmzfvzjH8d7Lr/88jhbVa3eC3bt2tVo7qijjqp//dd/jfe0vSe5zfPgxZ6rPoECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAoDOZ+yc7nc5TVbXqpXs4L6mDut3uvImGnPFlzxmf44wve874nF/UM06qQAGAZ/kKFwACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAI9kxmeNWtWd8GCBfGyvr6+OFtVtXHjxji7efPmevrppzsTzc2ePbu7ePHieM+0adPibFXVI488Emd37dpV4+PjE56xt7e32+ZvMTQ0FGerqtasWdMq3+12Jzxjp9PpdjoTjr2gAw88MM5WVW3atCnO7tq1q3bv3j3hg58yZUp36tSp8Z6fp71799a+ffsmPOOMGTO6bZ5v27Zti7NVVa985Svj7MjISI2Ojk54xsHBwe78+fPjPY8++micraoaHh6Os6Ojo7V9+/ZGz9WenknVzc844IAD4mxV1fr16+Ps7t27a+/evc97xkmdaMGCBXXdddfFD2TJkiVxtqrqpptuirN//dd/3Whu8eLF9c1vfjPe06Z8q6pOPPHEOPvv//7vjeb6+vrqyCOPjPecfvrpcbaq6k/+5E9a5ZvodDrV398f5z/60Y+22n/zzTfH2R/96EeN5qZOnVpz586N9+zbty/Ots1v2bKl0dzQ0FB94AMfiPfcd999cbaq6jvf+U6cXbZsWaO5+fPn1zXXXBPvOemkk+JsVdVll10WZz/2sY81muvp6Wn1XP3IRz4SZ6uqPvvZz8bZVatWveDPfIULAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgCBSV1ntnLlynrXu94VL/u93/u9OFvV7h7KpndD/vjHP251F+RZZ50VZ6uqdu7cGWebXi91wAEH1CWXXBLvufLKK+NsVdW3v/3tOPv+97+/0Vxvb2+rOwTPP//8OFtV1e1242zTa7Bmz55d73znO+M9ba54qvp/c8a1a9c2vjLr+fzgBz+Is1VVF198cZxteu/t9u3b67vf/W685w1veEOcrao65ZRT4uxVV13VaG727Nl19tlnx3vavh6/973vxdmnnnrqBX/mEygABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEJnUf6BFHHFHLly+Pl33uc5+Ls1VVF154Yat8E0uWLKnPf/7zcf7OO+9stf/GG2+Ms7/zO7/TaG7dunWt7vS866674mxV1f777x9n99tvv0ZzCxYsaHWX41FHHRVnq6q2bdsWZ/fu3dtobtOmTfUP//AP8Z7TTjstzlZVvf3tb4+zK1eubDTX7XZrfHw83nPvvffG2aqqBx54IM6OjY01mlu0aFF98pOfjPds2bIlzla1ez329DSrkPXr19cVV1wR73nd614XZ6uqjjvuuDj7YnfK+gQKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAgU63220+3Ok8VVWrXrqH85I6qNvtzptoyBlf9pzxOc74sueMz/lFPeOkChQAeJavcAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEg0DOZ4d7e3m5fX1+87MADD4yzVVX79u2Ls+vWrautW7d2Jprr6elpdcadO3fG2aqqbrfbNj/hGYeGhrqLFi2Kd2zcuDHOVlW12T0yMlKjo6MTnnHq1Kndnp5JPb1/xtKlS+NsVVWb3U3P2N/f391vv/3iPaOjo3G2qmrOnDlxdmxsrHbt2jXhGXt7e7v9/f3xnkMPPTTOVlVt3bo1zm7YsKHRe860adO6AwMD8Z4FCxbE2aqqNs+hps/Vvr6+7owZM+I98+bNi7NV7d6XN23aVGNjY897xkm9yvv6+urVr351/EA++9nPxtmqdr+E888/v9FcX19fvepVr4r3rFixIs5WtS/gJhYtWlRf/epX4/xNN93Uav8nPvGJOLts2bJGcz09Pa3eWL7zne/E2aqq2bNnx9mmZ9xvv/3qbW97W7znS1/6Upytqjr11FPj7D333NNorr+/v4488sh4zwMPPBBnq5o/zudz4YUXNpobGBio17/+9fGeiy++OM5WVf3Gb/xGnG36XJ0xY0addNJJ8Z4/+IM/iLNVVT/5yU/i7BVXXPGCP/MVLgAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEJnUby2GHHVY//OEP42WdzoS33ryoNrekrF69utFct9ut8fHxeM/dd98dZ6va3Vjz/e9/v9HcmjVr6pJLLon33H777XG2quqRRx6Js7t27Wo0N3/+/Fa3VJxxxhlxtqrddWZNfz+zZs2q3/7t3473/O3f/m2craq6+eab4+z999/faG737t21bt26eE/b95y1a9fG2enTpzeaO+igg+qLX/xivGdwcDDOVlVdcMEFcfbxxx9vNDd79uw688wz4z2//uu/Hmer2t0482J8AgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYDApC4tHBsba3yP30vhvPPOi7NN79l81ateVcuXL4/3nHjiiXG2qurqq6+Os29/+9sbzfX19dUhhxwS77nxxhvjbFXV+973vlb5JoaGhuptb3tbnP/2t7/dav/Xvva1VvkmfvrTn9Zpp50W5+fMmdNq/xe+8IU42/Sezr1799a2bdviPW3unq2qOuWUU+Lso48+2mhu48aNdf3118d7Pvaxj8XZqqpPf/rTcba3t7fx3AEHHBDv+fM///M4W1W1ffv2OPvGN77xBX/mEygABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEOt1ut/lwp/NUVa166R7OS+qgbrc7b6IhZ3zZc8bnOOPLnjM+5xf1jJMqUADgWb7CBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYBAz2SGZ8+e3V20aFG8rK+vL85WVW3fvj3Orlu3rrZu3dqZaG7q1Kndnp5J/Vp+xvj4eJytqmrz+928eXM9/fTTE55xcHCwO3/+/HjP2rVr42xV1cEHHxxnn3jiidq0adOEZ5w7d253eHg43vPoo4/G2aqq3t7eODs2Nla7du2a8IxDQ0OtXo8rVqyIs1VVBx54YJzdtGlTjY2NTXjGnp6ebpv3jZkzZ8bZqqrp06fH2dHR0dq+ffuEZ/x5v69u3bo1zm7YsKHR+2pPT0+3zWvi8MMPj7NtjYyM1Ojo6POecVJNsWjRovrHf/zH+IEsWbIkzlZVffvb346zf/iHf9horqenpxYsWBDvWb16dZytqvqjP/qjOHvNNdc0mps/f35dffXV8Z6Pf/zjcbaq6u///u/j7Fvf+tZGc8PDw7V8+fJ4z8knnxxnq579Hae+8Y1vNJpbtGhR3XnnnfGeQw89NM5WVV100UVx9sorr2w019fXV0uXLo33HH/88XG2quo1r3lNnL3ssssazS1atKi+/vWvx3sOOeSQOFtVdc8998TZCy+8sNFcb29vtfkHbZvXclXVvn374uwxxxzzgj/zFS4ABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABCZ1G0un06n+/v542aZNm+JsVdXKlSvj7DPPPNNo7rDDDqt777033rNw4cI4W1V1xRVXtMo38eSTT9anPvWpON/2ZoTTTjstzo6MjDSae+SRR+rEE0+M9/z+7/9+nK2quuSSS+Js02v7xsfHW93+84Y3vCHOVlV997vfjbPbtm1rNLd06dJWz7ctW7bE2aqqoaGhVvkmVq5cWWeeeWacb3vL1bvf/e5W+SamTp1ag4ODcf7hhx9utb9Nb73YFZU+gQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBgUveBTpkypdW9ameffXacrapW93Q2tWLFijr88MPj/Kmnntpqf5s7Fjds2NBo7tBDD2215z3veU+craq66aabWuWbGBoaqjPOOCPOv+Md72i1/+67746zF154YaO5ffv2Nb5X8/l8//vfj7NVVSeffHKc7Xa7jeY2bdpUt956a7zn6quvjrNV7V7PDzzwQKO5gYGBOuKII+I99913X5ytqrrtttta5ZsYHBxs9bs87LDDWu2/7rrr4uyLvcZ8AgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYBAp+m9fFVVnU7nqapa9dI9nJfUQd1ud95EQ874sueMz3HGlz1nfM4v6hknVaAAwLN8hQsAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAgZ7JDE+bNq07ffr0eNl+++0XZ6uq1qxZ0yrf7XY7E83MmDGjOzQ0FO944okn4mxV1Wtf+9o4OzIyUqOjoxOecdq0ad2BgYF4zy//8i/H2aqq8fHxOLtmzZratGnThGfsdDrdeElVtXkOVFWNjY3F2b1799bevXsnPOPcuXO7w8PD8Z7169fH2aqqKVPyf39v3LixxsbGJjxjX19fd8aMGfGeQw45JM5WVW3atCnOPvXUU7V9+/YJz9jT09Pt7e2N9xx++OFxtqrq0UcfjbM7d+6s8fHxCc84NDTUXbRoUbxnz549cbaqaubMmXH2xd5XJ1Wg06dPrze+8Y3xAznhhBPibFXVRRdd1CrfxNDQUH3gAx+I8x/5yEda7V++fHmcXbZsWaO5gYGBesMb3hDv+eY3vxlnq9r9Q+iUU05ptbup3/zN32yVf/DBB+PsunXrGs0NDw+3er5cddVVcbbq2edR6lOf+lSjuRkzZtRJJ50U77n11lvjbFXVzTffHGcvvfTSRnO9vb21ZMmSeE+b50BVu9fUD37wg0ZzixYtqjvvvDPes2HDhjhbVXXcccfF2WOPPfYFf+YrXAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACEzqOrNut1u7d++Ol+3YsSPOVrW7a/O3fuu3Gs09+eSTdfnll8d7rr766jhb1fyap+fT9Bqs8fHxevzxx+M9999/f5ytav63eD67du1qNHfUUUe1ulJszpw5cbaqqr+/P852u82uMl23bl19+tOfjvd8+MMfjrNV7e5M3bZtW6O5vr6+Vnd6Ll26NM5WVT300EOt8k0MDAzUYYcdFuf/4i/+otX+NtcTNr1Ccd++fY3/5s/n+OOPj7NV7d6XX+wqNZ9AASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIKBAASCgQAEgoEABIDCp+0CXLFlSd999d76sZ1Lr/i8nn3xynB0ZGWk0N2vWrHrLW94S75k3b16craq66KKL4mzTu1p3795da9asife86U1virNVVZdeemmc/eIXv9hortPptHq+vfrVr46zVVW33nprnD3llFMaze3YsaN+9KMfxXsWLVoUZ6uqVq1aFWePPfbYRnNbtmxp9Z7T19cXZ6uq1q5dG2ebvl8dfPDB9eUvfzneM3v27DhbVXXZZZe1yjexefPmuuOOO+J80ztyX8hZZ50VZ8fGxl7wZz6BAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQKAzmXvWOp3OU1WVXwL483VQt9ud8LJOZ3zZc8bnOOPLnjM+5xf1jJMqUADgWb7CBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYCAAgWAgAIFgIACBYBAz2SGp02b1h0YGIiXbd++Pc5WVS1YsCDObtmypXbs2NGZaG5oaKi7cOHCeM+2bdvibFXVmjVrWuW73e6EZ5wyZUp36tSp8Y49e/bE2f8vNDnj1KlTuz09k3p6/4w2z7WqqhkzZsTZtWvX1pYtWyY849y5c7vDw8PxnhUrVsTZqqq9e/fG2T179tS+ffsmPOO0adO6/f398Z42z4GqqrGxsTjb9IwDAwPdwcHBeM/s2bPjbFXVzp074+zo6Ght3759wjN2Op3ulCn557XDDjsszlZVtemtkZGRGh0dfd4zTurZNTAwUK973eviB/Ld7343zlZVnXfeeXH2S1/6UqO5hQsX1m233Rbvue++++JsVdXFF18cZ/ft29doburUqTV37tx4z+joaJyt+n9TwD09PbV48eI43+bvUFW1bNmyOPue97yn0dzw8HAtX7483rN06dI4W1W1devWONv0OdTf31/HHHNMvKfN87yq6v7774+zTc84ODhY55xzTrznne98Z5ytqvqv//qvOHvZZZc1mpsyZUq1+YfQLbfcEmerql7zmtfE2Rd7LfsKFwACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACk7qN5aCDDqq/+7u/i5edeuqpcbaq6hOf+ESrfBNPP/10Pfjgg3F+2rRprfY3vVGljTlz5tTZZ58d5++4445W+1euXBlnjz766EZzr371q1vdVNLpTHhD04v6p3/6pzjb9Laa1atX1wc/+MF4z0MPPRRnq6rOP//8OPvVr3610Vy3261nnnkm3nPrrbfG2ar2z4MmnnnmmXrsscfi/F/91V+12v/lL3+5Vb6JV77ylXXjjTfG+bPOOqvV/t27d8fZF7ti0idQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACEzqPtDx8fEXvRttIm3u9at69m7A1LJlyxrNzZ07t84777x4T5s776qq1e6vf/3rjeYWL15cV1xxRbxnxowZcbaqau/evXG26XNgw4YN9Td/8zfxnnPOOSfOVlVdcMEFcXb16tWN5nbs2NHqztObbropzlZVfeUrX4mz4+PjjeZ27dpVjzzySDViy9EAAAWASURBVLxn8+bNcbaqamBgIM7u2rWr0dyePXtq/fr18Z4zzjgjzlZV3XLLLXG26fvqypUrW93p2eYO4ap2dxj/6Z/+6Qv+zCdQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACHQmc8dmp9N5qqpWvXQP5yV1ULfbnTfRkDO+7Dnjc5zxZc8Zn/OLesZJFSgA8Cxf4QJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAoGcywzNnzuzOmTMnXjY6Ohpnq6p27NjRKt/tdjsTzfT29nb7+/vjHW2yVe1/R03OOHfu3O7w8HC849FHH42zVVW7d++Os+Pj47Vnz54Jz7j//vt3582bF+/Zb7/94mxV1Z49e+LsE088UZs3b57wjPvtt1937ty58Z5169bF2aqqffv2xdk9e/bU3r17X/Ln6n//93/H2ap2z4Nt27bVzp07Jzxjp9Ppxkuq6ogjjmgTr97e3jg7MjJSo6Ojjc44ZUr+eW3GjBlxtqqqzevkqaeequ3btz/vGSdVoHPmzKlLLrkkfiA33HBDnK2qWr58eat8E/39/XXkkUfG+cMOO6zV/uuuu65Vvonh4eFWv8tTTjml1f4nn3wyzj788MON5ubNm1ef/OQn4z1vfvOb42xV1fr16+Ps29/+9kZzc+fOrcsuuyze85nPfCbOVlU9/fTTcXbt2rWN5to+V3/1V381zlZVnXDCCXH2lltuabW7qfvuu69Vfv78+XF22bJljeamTJnS6sPFMcccE2erqs4777w4+2d/9mcv+DNf4QJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAQIECQECBAkBAgQJAYFK3sTzxxBMv+n+mn8j73//+OFtV9ZOf/CTOPvPMM43mfumXfqm+8IUvxHtuvPHGOFvV/JaK53PyySc3mvvf//3fOv300+M9K1asiLNVVatXr26Vb2LlypV11llnxfmvfOUrrfafccYZcbbprRWPP/54fehDH4r3bN26Nc5WtXs933HHHY3mHn/88brgggviPW1vnGlzg8i9997baO6oo46qBx98MN5z++23x9mqqiuvvDLOPvTQQ43mpk+fXkcddVS851vf+lacraq69dZbW+VfiE+gABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEFCgABBQoAAQUKAAEJjUfaBLliyp66+/Pl62cuXKOFtVtXPnzji7bNmyRnM//elP69RTT433tLnzrqrqgAMOiLO9vb2N5p555pl67LHH4j3nnHNOnK2q2rZtW5y97bbbGs295jWvqX/5l3+J9yxevDjOVlWtWrUqzq5fv77R3NKlS+uuu+6K9xx88MFxtqrd6/Hf/u3fGs3tv//+9eY3vznec9JJJ8XZqub3lj6f3bt3N5obGRmp9773vfGec889N85WVR1xxBFxtunzfHh4uK677rp4T6fTibNVz74fpNatW/eCP/MJFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAKdbrfbfLjTeaqq8osOf74O6na78yYacsaXPWd8jjO+7Dnjc35RzzipAgUAnuUrXAAIKFAACChQAAgoUAAIKFAACChQAAgoUAAIKFAACChQAAj8H3d4KmyrcEJ4AAAAAElFTkSuQmCC\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "conv_layer_index = [0,2,4]\n",
        "outputs = [model.layers[i].output for i in conv_layer_index]\n",
        "model_short = Model(inputs=model.inputs, outputs=outputs)\n",
        "print(model_short.summary())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IcyR9YMn7VO7",
        "outputId": "bb3c3582-60bb-4fa1-ec1c-186f61cb4f2b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv3d_input (InputLayer)   [(None, 22, 100, 100, 1)  0         \n",
            "                             ]                                   \n",
            "                                                                 \n",
            " conv3d (Conv3D)             (None, 20, 98, 98, 64)    1792      \n",
            "                                                                 \n",
            " max_pooling3d (MaxPooling3D  (None, 10, 49, 49, 64)   0         \n",
            " )                                                               \n",
            "                                                                 \n",
            " conv3d_1 (Conv3D)           (None, 8, 47, 47, 128)    221312    \n",
            "                                                                 \n",
            " max_pooling3d_1 (MaxPooling  (None, 4, 23, 23, 128)   0         \n",
            " 3D)                                                             \n",
            "                                                                 \n",
            " conv3d_2 (Conv3D)           (None, 3, 22, 22, 256)    262400    \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 485,504\n",
            "Trainable params: 485,504\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "columns = 8\n",
        "rows = 8\n",
        "for ftr in feature_output:\n",
        "    #pos = 1\n",
        "    fig=plt.figure(figsize=(12, 12))\n",
        "    for i in range(1, columns*rows +1):\n",
        "        fig =plt.subplot(rows, columns, i)\n",
        "        fig.set_xticks([])  #Turn off axis\n",
        "        fig.set_yticks([])\n",
        "        plt.imshow(ftr[0, :, :, i-1], cmap='gray')\n",
        "        #pos += 1\n",
        "    plt.show()"
      ],
      "metadata": {
        "id": "zQjqT4lJ8gBd"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}