gen_user_study_images.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Set CUDA:2 as current device.\n"
     ]
    }
   ],
   "source": [
    "import sys\n",
    "import os\n",
    "import torch\n",
    "import matplotlib.pyplot as plt\n",
    "import torchvision.transforms.functional as trans_f\n",
    "\n",
    "sys.path.append(os.path.abspath(sys.path[0] + '/../../'))\n",
    "__package__ = \"deep_view_syn.notebook\"\n",
    "torch.cuda.set_device(2)\n",
    "print(\"Set CUDA:%d as current device.\" % torch.cuda.current_device())\n",
    "\n",
    "from ..data.spherical_view_syn import *\n",
    "from ..configs.spherical_view_syn import SphericalViewSynConfig\n",
    "from ..my import netio\n",
    "from ..my import util\n",
    "from ..my import device\n",
    "from ..my import view\n",
    "from ..my.foveation import Foveation\n",
    "from ..my.gen_final import GenFinal\n",
    "\n",
    "\n",
    "def load_net(path):\n",
    "    config = SphericalViewSynConfig()\n",
    "    config.from_id(path[:-4])\n",
    "    config.SAMPLE_PARAMS['perturb_sample'] = False\n",
    "    config.print()\n",
    "    net = config.create_net().to(device.GetDevice())\n",
    "    netio.LoadNet(path, net)\n",
    "    return net\n",
    "\n",
    "\n",
    "def find_file(prefix):\n",
    "    for path in os.listdir():\n",
    "        if path.startswith(prefix):\n",
    "            return path\n",
    "    return None\n",
    "\n",
    "\n",
    "def load_views(data_desc_file) -> view.Trans:\n",
    "    with open(data_desc_file, 'r', encoding='utf-8') as file:\n",
    "        data_desc = json.loads(file.read())\n",
    "        samples = data_desc['samples'] if 'samples' in data_desc else [-1]\n",
    "        view_centers = torch.tensor(\n",
    "            data_desc['view_centers'], device=device.GetDevice()).view(samples + [3])\n",
    "        view_rots = torch.tensor(\n",
    "            data_desc['view_rots'], device=device.GetDevice()).view(samples + [3, 3])\n",
    "        return view.Trans(view_centers, view_rots)\n",
    "\n",
    "\n",
    "def plot_cross(center, res):\n",
    "    plt.plot(\n",
    "        [\n",
    "            (res[1] - 1) / 2 + center[0] - 5,\n",
    "            (res[1] - 1) / 2 + center[0] + 5\n",
    "        ],\n",
    "        [\n",
    "            (res[0] - 1) / 2 + center[1],\n",
    "            (res[0] - 1) / 2 + center[1]\n",
    "        ],\n",
    "        color=[0, 1, 0])\n",
    "    plt.plot(\n",
    "        [\n",
    "            (res[1] - 1) / 2 + center[0],\n",
    "            (res[1] - 1) / 2 + center[0]\n",
    "        ],\n",
    "        [\n",
    "            (res[0] - 1) / 2 + center[1] - 5,\n",
    "            (res[0] - 1) / 2 + center[1] + 5\n",
    "        ],\n",
    "        color=[0, 1, 0])\n",
    "\n",
    "\n",
    "def plot_figures(left_images, right_images, left_center, right_center):\n",
    "    # Plot Fovea raw\n",
    "    plt.figure(figsize=(8, 4))\n",
    "    plt.subplot(121)\n",
    "    util.PlotImageTensor(left_images['fovea_raw'])\n",
    "    plt.subplot(122)\n",
    "    util.PlotImageTensor(right_images['fovea_raw'])\n",
    "\n",
    "    # Plot Fovea\n",
    "    plt.figure(figsize=(8, 4))\n",
    "    plt.subplot(121)\n",
    "    util.PlotImageTensor(left_images['fovea'])\n",
    "    fovea_res = left_images['fovea'].size()[-2:]\n",
    "    plot_cross((0, 0), fovea_res)\n",
    "    plt.subplot(122)\n",
    "    util.PlotImageTensor(right_images['fovea'])\n",
    "    plot_cross((0, 0), fovea_res)\n",
    "\n",
    "    #plt.subplot(1, 4, 2)\n",
    "    # util.PlotImageTensor(fovea_refined)\n",
    "\n",
    "    # Plot Mid\n",
    "    plt.figure(figsize=(8, 4))\n",
    "    plt.subplot(121)\n",
    "    util.PlotImageTensor(left_images['mid'])\n",
    "    plt.subplot(122)\n",
    "    util.PlotImageTensor(right_images['mid'])\n",
    "\n",
    "    # Plot Periph\n",
    "    plt.figure(figsize=(8, 4))\n",
    "    plt.subplot(121)\n",
    "    util.PlotImageTensor(left_images['periph'])\n",
    "    plt.subplot(122)\n",
    "    util.PlotImageTensor(right_images['periph'])\n",
    "\n",
    "    # Plot Blended\n",
    "    plt.figure(figsize=(12, 6))\n",
    "    plt.subplot(121)\n",
    "    util.PlotImageTensor(left_images['blended'])\n",
    "    full_res = left_images['blended'].size()[-2:]\n",
    "    plot_cross(left_center, full_res)\n",
    "    plt.subplot(122)\n",
    "    util.PlotImageTensor(right_images['blended'])\n",
    "    plot_cross(right_center, full_res)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Change working directory to  /home/dengnc/deep_view_syn/data/__0_user_study/us_mc_all_in_one\n",
      "==== Config fovea ====\n",
      "Net type:  nmsl\n",
      "Encode dim:  10\n",
      "Optimizer decay:  0\n",
      "Normalize:  False\n",
      "Direction as input:  False\n",
      "Full-connected network parameters: {'nf': 128, 'n_layers': 4, 'skips': []}\n",
      "Sample parameters {'spherical': True, 'depth_range': (1.0, 50.0), 'n_samples': 32, 'perturb_sample': False, 'lindisp': True, 'inverse_r': True}\n",
      "==========================\n",
      "Load net from fovea@nmsl-rgb_e10_fc128x4_d1-50_s32.pth ...\n",
      "==== Config periph ====\n",
      "Net type:  nnmsl\n",
      "Encode dim:  10\n",
      "Optimizer decay:  0\n",
      "Normalize:  False\n",
      "Direction as input:  False\n",
      "Full-connected network parameters: {'nf': 64, 'n_layers': 4, 'skips': []}\n",
      "Sample parameters {'spherical': True, 'depth_range': (1.0, 50.0), 'n_samples': 16, 'perturb_sample': False, 'lindisp': True, 'inverse_r': True}\n",
      "==========================\n",
      "Load net from periph@nnmsl-rgb_e10_fc64x4_d1-50_s16.pth ...\n",
      "Dataset loaded.\n",
      "views: [5, 5, 5, 5, 5]\n"
     ]
    }
   ],
   "source": [
    "#os.chdir(sys.path[0] + '/../data/__0_user_study/us_gas_all_in_one')\n",
    "os.chdir(sys.path[0] + '/../data/__0_user_study/us_mc_all_in_one')\n",
    "#os.chdir(sys.path[0] + '/../data/bedroom_all_in_one')\n",
    "print('Change working directory to ', os.getcwd())\n",
    "torch.autograd.set_grad_enabled(False)\n",
    "\n",
    "fovea_net = load_net(find_file('fovea'))\n",
    "periph_net = load_net(find_file('periph'))\n",
    "\n",
    "# Load Dataset\n",
    "views = load_views('views.json')\n",
    "#ref_dataset = SphericalViewSynDataset('ref.json', load_images=False, calculate_rays=False)\n",
    "print('Dataset loaded.')\n",
    "\n",
    "print('views:', views.size())\n",
    "#print('ref views:', ref_dataset.samples)\n",
    "\n",
    "fov_list = [20, 45, 110]\n",
    "res_list = [(128, 128), (256, 256), (256, 230)]  # (192,256)]\n",
    "res_full = (1600, 1440)\n",
    "\n",
    "gen = GenFinal(fov_list, res_list, res_full, fovea_net, periph_net, device.GetDevice())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "view_coord: [2, 2, 2, 2, 2]\nshift: 3\nshift: -3\n"
     ]
    }
   ],
   "source": [
    "centers = [\n",
    "    # ==gas==\n",
    "    [(-137, 64), (-142, 64)],\n",
    "    [(133, -44), (130, -44)],\n",
    "    [(-20, -5), (-25, -5)],\n",
    "    # ==mc==\n",
    "    [(-107, 80), (-112, 80)],\n",
    "    [(-17, -90), (-22, -90)],\n",
    "    [(95, 30), (91, 30)]\n",
    "]\n",
    "set_id = 5\n",
    "\n",
    "view_coord = [0, 0, 0, 0, 0]\n",
    "for i, val in enumerate(views.size()):\n",
    "    view_coord[i] += val // 2\n",
    "print('view_coord:', view_coord)\n",
    "test_view = views.get(*view_coord)\n",
    "\n",
    "left_images = gen(centers[set_id][0], view.Trans(\n",
    "    test_view.trans_point(\n",
    "        torch.tensor([-0.03, 0, 0], device=device.GetDevice())\n",
    "    ), test_view.r), mono_trans=test_view, ret_raw=True)\n",
    "right_images = gen(centers[set_id][1], view.Trans(\n",
    "    test_view.trans_point(\n",
    "        torch.tensor([0.03, 0, 0], device=device.GetDevice())\n",
    "    ), test_view.r), mono_trans=test_view, ret_raw=True)\n",
    "\n",
    "#plot_figures(left_images, right_images, centers[set_id][0], centers[set_id][1])\n",
    "\n",
    "util.CreateDirIfNeed('output')\n",
    "for key in left_images:\n",
    "    util.WriteImageTensor(\n",
    "        left_images[key], 'output/set%d_%s_l.png' % (set_id, key))\n",
    "for key in right_images:\n",
    "    util.WriteImageTensor(\n",
    "        right_images[key], 'output/set%d_%s_r.png' % (set_id, key))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3.7.9 64-bit ('pytorch': conda)",
   "metadata": {
    "interpreter": {
     "hash": "660ca2a75467d3af74a68fcc6f40bc78ab96b99ff17d2f100b5ca821fbb183f2"
    }
   }
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9-final"
  },
  "orig_nbformat": 2
 },
 "nbformat": 4,
 "nbformat_minor": 2
}