model_training.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "Hand-Recog.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.6.4"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "24FvuPVCoq8X"
      },
      "source": [
        "Link To Dataset : https://www.kaggle.com/sarjit07/hand-gesture-recog-dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EAZqytJbaoFP",
        "outputId": "7f0f2311-01d6-4d78-b224-ea4155ec7f8b"
      },
      "source": [
        "!nvidia-smi"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Thu Dec  3 07:07:51 2020       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 455.38       Driver Version: 418.67       CUDA Version: 10.1     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   46C    P8     9W /  70W |      0MiB / 15079MiB |      0%      Default |\n",
            "|                               |                      |                 ERR! |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LavJhhK703Pj"
      },
      "source": [
        "from google.colab import files\n",
        "\n",
        "import warnings\n",
        "warnings.simplefilter(\"ignore\")"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 91
        },
        "id": "wNIYjAM3YH-Y",
        "outputId": "e340bd8e-ea70-4495-8955-cbbc36ad1269"
      },
      "source": [
        "!pip install -q kaggle\n",
        "files.upload()"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-cbe48b78-1e92-460b-9a16-c1891244f6a3\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-cbe48b78-1e92-460b-9a16-c1891244f6a3\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Saving kaggle.json to kaggle.json\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'kaggle.json': b'{\"username\":\"synysterjeet\",\"key\":\"5137575186c9d4f987e712fff55adc74\"}'}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gSsqqKzJYKJB"
      },
      "source": [
        "!mkdir ~/.kaggle\n",
        "!cp kaggle.json ~/.kaggle\n",
        "!chmod 600 ~/.kaggle/kaggle.json"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CmiG08cnYYKg",
        "outputId": "7c568732-c870-4110-f621-9accaca52dfa"
      },
      "source": [
        "!kaggle datasets download -d sarjit07/hand-gesture-recog-dataset"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading hand-gesture-recog-dataset.zip to /content\n",
            " 58% 25.0M/42.9M [00:01<00:00, 24.1MB/s]\n",
            "100% 42.9M/42.9M [00:01<00:00, 38.4MB/s]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LWYTAOax1ESi"
      },
      "source": [
        "!unzip -q hand-gesture-recog-dataset.zip -d HandGest"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I8rlxwiq0s7B"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import os\n",
        "import tensorflow as tf\n",
        "import cv2\n",
        "import random\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras.layers import Dense, Input, InputLayer, Flatten\n",
        "from tensorflow.keras import models, layers\n",
        "from tensorflow.keras.models import Sequential, Model\n",
        "from  matplotlib import pyplot as plt\n",
        "import matplotlib.image as mpimg\n",
        "%matplotlib inline"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 228
        },
        "id": "TN5RR8HJ0s7C",
        "outputId": "3cfdce11-c5ae-4433-b036-9c6b5cd7f71d"
      },
      "source": [
        "plt.figure(figsize=(20,20))\n",
        "img_folder=\"HandGest/data/thumbsup\"\n",
        "for i in range(5):\n",
        "    file = random.choice(os.listdir(img_folder))\n",
        "    image_path= os.path.join(img_folder, file)\n",
        "    img=mpimg.imread(image_path)\n",
        "    ax=plt.subplot(1,5,i+1)\n",
        "    ax.title.set_text(file)\n",
        "    plt.imshow(img)"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x1440 with 5 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RnaJbHyuqH4B"
      },
      "source": [
        "IMG_WIDTH=125\n",
        "IMG_HEIGHT=125\n",
        "img_folder='HandGest/data'"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zxrl0d0ob2UL"
      },
      "source": [
        "!rm -r 'HandGest/data/blank'"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yy58c8qAqNQx"
      },
      "source": [
        "def create_dataset(img_folder):\n",
        "   \n",
        "    img_data_array=[]\n",
        "    class_name=[]\n",
        "   \n",
        "    for dir1 in os.listdir(img_folder):\n",
        "        for file in os.listdir(os.path.join(img_folder, dir1)):\n",
        "       \n",
        "            image_path= os.path.join(img_folder, dir1,  file)\n",
        "            image= cv2.imread( image_path, cv2.COLOR_BGR2RGB)\n",
        "            image=cv2.resize(image, (IMG_HEIGHT, IMG_WIDTH),interpolation = cv2.INTER_AREA)\n",
        "            image=np.array(image)\n",
        "            image = image.astype('float32')\n",
        "            image /= 255 \n",
        "            image = np.stack((image,)*3, axis=-1)\n",
        "            img_data_array.append(image)\n",
        "            class_name.append(dir1)\n",
        "    return img_data_array, class_name\n",
        "    \n",
        "# extract the image array and class name\n",
        "img_data, class_name =create_dataset(img_folder)"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JIY_xJfxqUb8",
        "outputId": "fabb68e9-975e-422c-d636-e6fe5619ef84"
      },
      "source": [
        "target_dict={k: v for v, k in enumerate(np.unique(class_name))}\n",
        "print(target_dict)\n",
        "target_val= [target_dict[class_name[i]] for i in range(len(class_name))]"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "{'fist': 0, 'five': 1, 'ok': 2, 'thumbsdown': 3, 'thumbsup': 4}\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TXWRVvm-qrnK",
        "outputId": "438e658c-7a16-460b-b857-7d61f1d3f7f1"
      },
      "source": [
        "x_data = np.array(img_data, np.float32)\n",
        "y_data = np.array(list(map(int,target_val)))\n",
        "y_data = y_data.reshape(y_data.shape[0],1)\n",
        "print(x_data.shape, y_data.shape)"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(8705, 125, 125, 3) (8705, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UVdS3hzotHCA"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "x_train,x_further,y_train,y_further = train_test_split(x_data,y_data,test_size = 0.2)\n",
        "x_validate,x_test,y_validate,y_test = train_test_split(x_further,y_further,test_size = 0.5)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rGqxwM_Vqk1f",
        "outputId": "7ff87534-e977-4a2a-ffb6-f21973e69072"
      },
      "source": [
        "model=models.Sequential()\n",
        "model.add(layers.Conv2D(32, (5, 5), strides=(2, 2), activation='relu', input_shape=(125, 125,3))) \n",
        "model.add(layers.MaxPooling2D((2, 2)))\n",
        "model.add(layers.Conv2D(64, (3, 3), activation='relu')) \n",
        "model.add(layers.MaxPooling2D((2, 2)))\n",
        "model.add(layers.Flatten())\n",
        "model.add(layers.Dense(5, activation='softmax'))\n",
        "model.compile(optimizer='rmsprop', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
        "model.summary()"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_2 (Conv2D)            (None, 61, 61, 32)        2432      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 30, 30, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 28, 28, 64)        18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 14, 14, 64)        0         \n",
            "_________________________________________________________________\n",
            "flatten_2 (Flatten)          (None, 12544)             0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 5)                 62725     \n",
            "=================================================================\n",
            "Total params: 83,653\n",
            "Trainable params: 83,653\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ci6yLxVJrzco",
        "outputId": "01bf9572-4e45-4c92-be0e-6376de68abc2"
      },
      "source": [
        "history = model.fit(x_train, y_train, epochs=10, batch_size=16, verbose=1, validation_data=(x_validate, y_validate))"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "436/436 [==============================] - 2s 5ms/step - loss: 0.1054 - accuracy: 0.9614 - val_loss: 0.0035 - val_accuracy: 0.9989\n",
            "Epoch 2/10\n",
            "436/436 [==============================] - 2s 5ms/step - loss: 0.0059 - accuracy: 0.9989 - val_loss: 2.5672e-05 - val_accuracy: 1.0000\n",
            "Epoch 3/10\n",
            "436/436 [==============================] - 2s 4ms/step - loss: 0.0021 - accuracy: 0.9991 - val_loss: 2.5360e-06 - val_accuracy: 1.0000\n",
            "Epoch 4/10\n",
            "436/436 [==============================] - 2s 4ms/step - loss: 7.0891e-04 - accuracy: 0.9999 - val_loss: 2.0653e-06 - val_accuracy: 1.0000\n",
            "Epoch 5/10\n",
            "436/436 [==============================] - 2s 4ms/step - loss: 0.0043 - accuracy: 0.9993 - val_loss: 1.3812e-07 - val_accuracy: 1.0000\n",
            "Epoch 6/10\n",
            "436/436 [==============================] - 2s 4ms/step - loss: 0.0018 - accuracy: 0.9996 - val_loss: 1.4875e-06 - val_accuracy: 1.0000\n",
            "Epoch 7/10\n",
            "436/436 [==============================] - 2s 5ms/step - loss: 7.1437e-08 - accuracy: 1.0000 - val_loss: 5.0698e-09 - val_accuracy: 1.0000\n",
            "Epoch 8/10\n",
            "436/436 [==============================] - 2s 4ms/step - loss: 1.1983e-10 - accuracy: 1.0000 - val_loss: 8.6324e-09 - val_accuracy: 1.0000\n",
            "Epoch 9/10\n",
            "436/436 [==============================] - 2s 4ms/step - loss: 3.4236e-11 - accuracy: 1.0000 - val_loss: 5.8919e-09 - val_accuracy: 1.0000\n",
            "Epoch 10/10\n",
            "436/436 [==============================] - 2s 5ms/step - loss: 1.7118e-11 - accuracy: 1.0000 - val_loss: 4.2477e-09 - val_accuracy: 1.0000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4lhDyUTV0s7C",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "88bacecf-27ed-4700-8c81-2dfee46cc912"
      },
      "source": [
        "[loss, acc] = model.evaluate(x_test,y_test,verbose=1)\n",
        "print(\"Accuracy:\" + str(acc))"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "28/28 [==============================] - 0s 5ms/step - loss: 1.9161e-09 - accuracy: 1.0000\n",
            "Accuracy:1.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "collapsed": true,
        "id": "JXEgilWA0s7C"
      },
      "source": [
        "You'll get slightly different numbers each time you run it but you should be getting between 99.9 and 100% accuracy. Great!"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MYEk5MVbjSjG",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ba5a23ed-86ad-4bd7-d00c-ca237f77028e"
      },
      "source": [
        "image = cv2.imread(\"/content/HandGest/data/thumbsdown/down0.jpg\")\n",
        "image = cv2.resize(image, (125,125))\n",
        "image=np.array(image)\n",
        "image = image.astype('float32')\n",
        "image /= 255 \n",
        "image = image.reshape(-1,125,125,3)\n",
        "image.shape\n",
        "np.argmax(model.predict(image), axis=-1)[0]"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "3"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bbHRbswewuag"
      },
      "source": [
        "model.save(\"CNN.h5\")"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3Lq4Vv_cc_Ae"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}