SVM_classification_with_support_vectors.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "SVM_classification_with_support_vectors.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "R2Qx46DeFsAd",
        "colab_type": "code",
        "outputId": "c3d8437e-957e-4386-f29c-95ccdb9990d0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 546
        }
      },
      "source": [
        "import cv2 as cv\n",
        "import numpy as np\n",
        "from google.colab.patches import cv2_imshow\n",
        "\n",
        "# Set up training data\n",
        "# (1) basic example:C=1\n",
        "labels = np.array([1, -1, -1, -1])\n",
        "trainingData = np.matrix([[501, 10], [255, 10], [501, 255], [10, 501]], dtype=np.float32)\n",
        "\n",
        "# (2) 4-point example:C=1\n",
        "#labels = np.array([1, 1, -1, -1])\n",
        "#trainingData = np.matrix([[300, 300], [400, 301], [300, 500], [400, 499]], dtype=np.float32)\n",
        "\n",
        "# (3) complicated example: please try to set C=0.00001 & C=1\n",
        "#labels = np.array([1, 1, 1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1])\n",
        "#trainingData = np.matrix([[100, 100], [101, 200], [100, 300], [150, 200], [400, 50], [400, 100], [400, 200], [400,300], [400, 350], [500, 50], [500, 100], [500, 200], [500,300], [500, 350]], dtype=np.float32)\n",
        "\n",
        "\n",
        "# Train the SVM\n",
        "svm = cv.ml.SVM_create()\n",
        "svm.setType(cv.ml.SVM_C_SVC)\n",
        "svm.setKernel(cv.ml.SVM_LINEAR)\n",
        "svm.setTermCriteria((cv.TERM_CRITERIA_MAX_ITER, 100, 1e-6))\n",
        "#svm.setC(0.00001)\n",
        "svm.setC(1)\n",
        "\n",
        "svm.train(trainingData, cv.ml.ROW_SAMPLE, labels)\n",
        "\n",
        "# Data for visual representation\n",
        "width = 512\n",
        "height = 512\n",
        "image = np.zeros((height, width, 3), dtype=np.uint8)\n",
        "# Show the decision regions given by the SVM\n",
        "green = (0,255,0)\n",
        "blue = (255,0,0)\n",
        "for i in range(image.shape[0]):\n",
        "    for j in range(image.shape[1]):\n",
        "        sampleMat = np.matrix([[j,i]], dtype=np.float32)\n",
        "        response = svm.predict(sampleMat)[1]\n",
        "        if response == 1:\n",
        "            image[i,j] = green\n",
        "        elif response == -1:\n",
        "            image[i,j] = blue\n",
        "# Show the training data\n",
        "thickness = -1\n",
        "\n",
        "h, w = trainingData.shape\n",
        "\n",
        "for i in range(h):\n",
        "  if labels[i] == 1:\n",
        "    cv.circle(image, (trainingData[i,0],  trainingData[i,1]), 5, (  0,   0,   0), thickness)\n",
        "  else:\n",
        "    cv.circle(image, (trainingData[i,0],  trainingData[i,1]), 5, (  255,   255,   25), thickness)\n",
        "\n",
        "\n",
        "# Show support vectors\n",
        "thickness = 2\n",
        "sv = svm.getUncompressedSupportVectors()\n",
        "for i in range(sv.shape[0]):\n",
        "    cv.circle(image, (sv[i,0], sv[i,1]), 6, (128, 128, 128), thickness)\n",
        "#cv.imwrite('result.png', image) # save the image\n",
        "cv2_imshow(image) # show it to the user\n",
        "\n",
        "print('the number of support vectors:{}'.format(sv.shape[0]))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgAAAAIACAIAAAB7GkOtAAAKWklEQVR4nO3dy1nj2BpAUdHfTaIj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\nAIgSAICo/wDwZUEK8J+3RgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<PIL.Image.Image image mode=RGB size=512x512 at 0x7F4EFF521470>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "the number of support vectors:3\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e7A_8y0UE4wj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "3"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}