train_only.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Im4I9VxYa5Sw"
   },
   "source": [
    "<center><img src=\"./images/DLI Header.png\" alt=\"Header\" style=\"width: 400px;\"/></center>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "eI7xuIu1a5TA"
   },
   "source": [
    "<h1 style=\"background-color:#76b900;\"></h1>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# DLI Ambassador Project - Part II\n",
    "\n",
    "### Interactive Clock Reader - Training with collected data only, No live camera"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install livelossplot --quiet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "name in DATASETS:  A\n",
      "datasets:  <dataset.ImageClassificationDataset object at 0x7f58572b3d68>\n",
      "clock_fingers task with ['1', '2', '3', '4', '5', '6', '7', '8', '9', '10', '11', '12'] categories defined\n"
     ]
    }
   ],
   "source": [
    "import torchvision.transforms as transforms\n",
    "from dataset import ImageClassificationDataset\n",
    "\n",
    "TASK = 'clock_fingers'\n",
    "# TASK = 'thumbs'\n",
    "# TASK = 'emotions'\n",
    "# TASK = 'fingers'\n",
    "# TASK = 'diy'\n",
    "\n",
    "CATEGORIES = ['1', '2', '3', '4', '5', '6','7','8','9','10','11','12']\n",
    "# CATEGORIES = ['thumbs_up', 'thumbs_down']\n",
    "# CATEGORIES = ['none', 'happy', 'sad', 'angry']\n",
    "# CATEGORIES = ['1', '2', '3', '4', '5']\n",
    "# CATEGORIES = [ 'diy_1', 'diy_2', 'diy_3']\n",
    "\n",
    "DATASETS = ['A']\n",
    "# DATASETS = ['A', 'B', 'C']\n",
    "# DATASETS = ['A', 'B']\n",
    "\n",
    "TRANSFORMS = transforms.Compose([\n",
    "    transforms.ColorJitter(0.2, 0.2, 0.2, 0.2),\n",
    "    transforms.Resize((224, 224)),\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "])\n",
    "\n",
    "\n",
    "# Set up the data directory location if not there already\n",
    "#DATA_DIR = '/nvdli-nano/data/clock_finger_classification/'\n",
    "DATA_DIR = '/home/joosm/Workplace/DLI_ambassador/data/'\n",
    "!mkdir -p {DATA_DIR}\n",
    "\n",
    "datasets = {}\n",
    "for name in DATASETS:\n",
    "    print(\"name in DATASETS: \", name)\n",
    "    datasets[name] = ImageClassificationDataset(DATA_DIR + TASK + '_' + name, CATEGORIES, TRANSFORMS)\n",
    "    print(\"datasets: \",datasets[name])\n",
    "    \n",
    "print(\"{} task with {} categories defined\".format(TASK, CATEGORIES))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchvision\n",
    "\n",
    "\n",
    "device = torch.device('cuda')\n",
    "\n",
    "# ALEXNET\n",
    "# model = torchvision.models.alexnet(pretrained=True)\n",
    "# model.classifier[-1] = torch.nn.Linear(4096, len(dataset.categories))\n",
    "\n",
    "# SQUEEZENET \n",
    "# model = torchvision.models.squeezenet1_1(pretrained=True)\n",
    "# model.classifier[1] = torch.nn.Conv2d(512, len(dataset.categories), kernel_size=1)\n",
    "# model.num_classes = len(dataset.categories)\n",
    "\n",
    "# RESNET 18\n",
    "model = torchvision.models.resnet18(pretrained=True)\n",
    "model.fc = torch.nn.Linear(512, len(dataset.categories))\n",
    "\n",
    "# RESNET 34\n",
    "# model = torchvision.models.resnet34(pretrained=True)\n",
    "# model.fc = torch.nn.Linear(512, len(dataset.categories))\n",
    "    \n",
    "model = model.to(device)\n",
    "\n",
    "model_save_button = ipywidgets.Button(description='save model')\n",
    "#model_load_button = ipywidgets.Button(description='load model')\n",
    "MODEL_DIR = '/home/joosm/Workplace/DLI_ambassador/data/model_epoch05/'\n",
    "#model_path_widget = ipywidgets.Text(description='model path', value = '/nvdli-nano/data/clock_finger_classification/my_model.pth')\n",
    "model_path_widget = ipywidgets.Text(description='model path', value = MODEL_DIR + 'model.pth')\n",
    "\n",
    "   \n",
    "def save_model(c):\n",
    "    torch.save(model.state_dict(), model_path_widget.value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy\n",
      "\taccuracy         \t (min:    0.697, max:    0.995, cur:    0.958)\n",
      "Loss\n",
      "\tloss             \t (min:    0.004, max:    0.119, cur:    0.016)\n"
     ]
    }
   ],
   "source": [
    "from livelossplot import PlotLosses\n",
    "#from livelossplot.outputs import MatplotlibPlot\n",
    "#import PIL.Image\n",
    "\n",
    "BATCH_SIZE = 8\n",
    "num_epoch = 10\n",
    "\n",
    "optimizer = torch.optim.Adam(model.parameters())\n",
    "# optimizer = torch.optim.SGD(model.parameters(), lr=1e-3, momentum=0.9)\n",
    "\n",
    "\n",
    "def train_eval(is_training, num_epoch):\n",
    "    liveloss = PlotLosses()\n",
    "    global BATCH_SIZE, LEARNING_RATE, MOMENTUM, model, dataset, optimizer, train_button, accuracy_widget, loss_widget, progress_widget\n",
    "    \n",
    "    try:\n",
    "        train_loader = torch.utils.data.DataLoader(\n",
    "            dataset,\n",
    "            batch_size=BATCH_SIZE,\n",
    "            shuffle=True\n",
    "        )\n",
    "\n",
    "\n",
    "        time.sleep(1)\n",
    "\n",
    "        if is_training:\n",
    "            model = model.train()\n",
    "        else:\n",
    "            print(\"training mode error\")\n",
    "            break\n",
    "\n",
    "        while num_epoch > 0:            \n",
    "            logs = {}\n",
    "            i = 0\n",
    "            sum_loss = 0.0\n",
    "            error_count = 0.0\n",
    "            for images, labels in iter(train_loader):\n",
    "                # send data to device\n",
    "                images = images.to(device)\n",
    "                labels = labels.to(device)\n",
    "\n",
    "                if is_training:\n",
    "                    # zero gradients of parameters\n",
    "                    optimizer.zero_grad()\n",
    "\n",
    "                # execute model to get outputs\n",
    "                outputs = model(images)\n",
    "\n",
    "                # compute loss\n",
    "                loss = F.cross_entropy(outputs, labels)\n",
    "\n",
    "                if is_training:\n",
    "                    # run backpropogation to accumulate gradients\n",
    "                    loss.backward()\n",
    "\n",
    "                    # step optimizer to adjust parameters\n",
    "                    optimizer.step()\n",
    "\n",
    "                # increment progress\n",
    "                error_count += len(torch.nonzero(outputs.argmax(1) - labels).flatten())\n",
    "                count = len(labels.flatten())\n",
    "                i += count\n",
    "                sum_loss += float(loss)\n",
    "\n",
    "            logs['loss'] = sum_loss / i  \n",
    "            logs['accuracy'] = 1.0 - error_count / i\n",
    "            liveloss.update(logs)    \n",
    "\n",
    "            liveloss.send()\n",
    "            if is_training:\n",
    "                num_epoch = num_epoch - 1\n",
    "            else:\n",
    "                break\n",
    "    except e:\n",
    "        pass\n",
    "    \n",
    "\n",
    "train_eval(is_training=True, num_epoch = num_epoch)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1 style=\"background-color:#76b900;\"></h1>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4NwiSUhya5TA"
   },
   "source": [
    "<center><img src=\"./images/DLI Header.png\" alt=\"Header\" style=\"width: 400px;\"/></center>"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "classification_interactive.ipynb",
   "provenance": []
  },
  "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}