Depth demonising experiments with TV.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import open3d as o3d\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "#import OpenEXR as exr, Imath\n",
    "from tqdm import tqdm\n",
    "from glob import glob\n",
    "import cv2\n",
    "\n",
    "import torch\n",
    "import torch as th\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "\n",
    "from utils.cloud import Cloud\n",
    "from pose_est import PoseEst, create_rot_from_angle, transform_pc, extrinsics_from_rot_transl, get_params_from_rot_and_transl\n",
    "from params_est import optim_step_ICP, optim_step_neighbors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "depth_dir = f'../data/dpth/'\n",
    "gt_extrinsics_dir = f'../data/gt_dpth/'\n",
    "gt_depth_dir = '../data/gt_dpth/'\n",
    "shared_K_path = f'../data/gt_dpth/K_0.txt'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Testing denoising"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Input data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_data(depth_dir, gt_extrinsics_dir, indices, gt_depth_dir=None):\n",
    "    # indices: array-like of int\n",
    "    depth_all, RT_all, gt_depth_all = [], [], []\n",
    "    for ind in indices:\n",
    "        depth = np.load(os.path.join(depth_dir, f'{ind}.npy'))\n",
    "        RT = np.loadtxt(os.path.join(gt_extrinsics_dir, f'RT_{ind}.txt'))\n",
    "        depth_all.append(depth)\n",
    "        RT_all.append(RT)\n",
    "        if gt_depth_dir is not None:\n",
    "            gt_depth = np.load(os.path.join(gt_depth_dir, f'{ind}.npy'))\n",
    "            if len(gt_depth.shape) == 3 and gt_depth.shape[2] == 1:\n",
    "                gt_depth = gt_depth[..., 0]\n",
    "            gt_depth_all.append(gt_depth)\n",
    "    \n",
    "    depth_all = torch.tensor(np.stack(depth_all)).double()\n",
    "    RT_all = torch.tensor(np.stack(RT_all)).double()\n",
    "    if gt_depth_dir is not None:\n",
    "        gt_depth_all = torch.tensor(np.stack(gt_depth_all)).double()\n",
    "        return depth_all, RT_all, gt_depth_all\n",
    "    return depth_all, RT_all"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "depth, extr, gt_depth = load_data(depth_dir, gt_extrinsics_dir, indices=[80, 90], gt_depth_dir=gt_depth_dir)\n",
    "depth = depth.cuda()\n",
    "extr = extr.cuda()\n",
    "gt_depth = gt_depth.cuda()\n",
    "\n",
    "depth = F.interpolate(depth.unsqueeze(0), scale_factor=1).squeeze()\n",
    "gt_depth = F.interpolate(gt_depth.unsqueeze(0), scale_factor=1).squeeze()\n",
    "\n",
    "d1 = gt_depth[0].clone()\n",
    "d2 = gt_depth[1].clone()\n",
    "d1[d1 != d1.max()] += (th.rand_like(d1[d1 != d1.max()]) - 0.5)*2/10*d1[d1 != d1.max()]\n",
    "\n",
    "pc0 = Cloud.from_tensors(gt_depth[0].cuda(), K.cuda(), torch.eye(4).cuda(), ignore='max')\n",
    "pc1 = Cloud.from_tensors(gt_depth[1].cuda(), K.cuda(), torch.eye(4).cuda(), ignore='max')\n",
    "\n",
    "pc0_noisy = Cloud.from_tensors(d1.cuda(), K.cuda(), torch.eye(4).cuda(), ignore='max')\n",
    "pc1_noisy = Cloud.from_tensors(d2.cuda(), K.cuda(), torch.eye(4).cuda(), ignore='max')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "def loss_T(T_estim, extr_first, extr_second):\n",
    "    gt_rel_extr = extr_second @ torch.inverse(extr_first)\n",
    "    loss_val = ((gt_rel_extr - T_estim) ** 2).sum()\n",
    "    return loss_val\n",
    "\n",
    "def loss_params(pred_params, extr_first, extr_second):\n",
    "    gt_rel_extr = extr_second @ torch.inverse(extr_first)\n",
    "    gt_params = get_params_from_rot_and_transl(gt_rel_extr[:3, :3], gt_rel_extr[:3, 3])\n",
    "    loss_val = ((pred_params - gt_params) ** 2).sum()\n",
    "    return loss_val\n",
    "\n",
    "def grad_depth(dpth):\n",
    "    g_filters = th.zeros(2, 1, 3, 3).to(dpth)\n",
    "    g_filters[0, 0, 0, 1] = 1\n",
    "    g_filters[0, 0, 1, 1] = -1\n",
    "    g_filters[1, 0, 1, 0] = 1\n",
    "    g_filters[1, 0, 1, 1] = -1\n",
    "    ret = F.conv2d(dpth[None, None, :, :], g_filters, padding=1).squeeze()\n",
    "    return ret\n",
    "\n",
    "def l1_depth(dpth):\n",
    "    mask = dpth != dpth.max()\n",
    "    mask = mask.unsqueeze(0).expand(2, -1, -1)\n",
    "    grad = grad_depth(dpth)\n",
    "    norm = th.norm(grad[mask], p=1)\n",
    "    return norm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "def perform_gradslam(pc0, pc1, gradslam_iter=10):\n",
    "    T = torch.eye(4).double().cuda()\n",
    "    for i in (range(gradslam_iter)):\n",
    "        params_upd = optim_step_neighbors(pc0, pc1,\n",
    "                                          lam_min=0.00001,\n",
    "                                          lam_max=0.01,\n",
    "                                          D=1,\n",
    "                                          sigma=1e-4,\n",
    "                                          max_iter=20,\n",
    "                                          neighbors_take_each=1000, verbose=0)\n",
    "        transl_upd, angle_upd = params_upd.split([3, 3], dim=-1)\n",
    "        rot_upd = create_rot_from_angle(angle_upd.unsqueeze(0)).squeeze(0)\n",
    "        pc0.points = transform_pc(pc0.points, rot_upd, transl_upd).squeeze(1)\n",
    "\n",
    "        T_upd = extrinsics_from_rot_transl(rot_upd, transl_upd).double()\n",
    "        T = T_upd @ T\n",
    "        rot, transl = T[:3, :3], T[:3, 3]\n",
    "        params = get_params_from_rot_and_transl(rot, transl)\n",
    "        #print('loss', loss_params(params, extr[0], extr[1]))\n",
    "    return params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss_denoise =  0.011741241878238435\n",
      "torch.Size([147518, 3])\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Loss_align =  0.0030242917487292693\n",
      "Loss_TV =  1.669327377014578e-05\n",
      "Loss_denoise =  0.01172846988506117\n",
      "torch.Size([147518, 3])\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Loss_align =  9.63810892589593\n",
      "Loss_TV =  1.738118894134774e-05\n",
      "Loss_denoise =  0.011715783154860404\n",
      "torch.Size([147518, 3])\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Loss_align =  33.216420995934634\n",
      "Loss_TV =  1.76074551831214e-05\n",
      "Loss_denoise =  0.011703145079092264\n",
      "torch.Size([147518, 3])\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Loss_align =  19.08144080167834\n",
      "Loss_TV =  1.7722988582182513e-05\n",
      "Loss_denoise =  0.01169058001413658\n",
      "torch.Size([147518, 3])\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Loss_align =  31.40959540194246\n",
      "Loss_TV =  1.7811542600431118e-05\n",
      "Loss_denoise =  0.01167806736505396\n",
      "torch.Size([147518, 3])\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n",
      "Did not converge with required tolerance.\n"
     ]
    },
    {
     "ename": "RuntimeError",
     "evalue": "CUDA out of memory. Tried to allocate 2.28 GiB (GPU 0; 10.92 GiB total capacity; 6.17 GiB already allocated; 1.83 GiB free; 2.34 GiB cached)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-33-6ad06f10da7a>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     10\u001b[0m     \u001b[0mpc1_noisy\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mCloud\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfrom_tensors\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0md2\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcuda\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mK\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcuda\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0meye\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m4\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcuda\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mignore\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'max'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m     \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpc0\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpoints\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 12\u001b[0;31m     \u001b[0mparams\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mperform_gradslam\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpc0_noisy\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpc1_noisy\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mgradslam_iter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m15\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     13\u001b[0m     \u001b[0mloss_val\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mloss_params\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mparams\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mextr\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mextr\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0mloss_tv\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0ml1_depth\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0md1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mlam_norm\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-27-4bea3dd86b21>\u001b[0m in \u001b[0;36mperform_gradslam\u001b[0;34m(pc0, pc1, gradslam_iter)\u001b[0m\n\u001b[1;32m      8\u001b[0m                                           \u001b[0msigma\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1e-4\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m                                           \u001b[0mmax_iter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m20\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m                                           neighbors_take_each=1000, verbose=0)\n\u001b[0m\u001b[1;32m     11\u001b[0m         \u001b[0mtransl_upd\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mangle_upd\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mparams_upd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m3\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdim\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m         \u001b[0mrot_upd\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcreate_rot_from_angle\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mangle_upd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munsqueeze\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msqueeze\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/gradSLAM_FinalProject/params_est.py\u001b[0m in \u001b[0;36moptim_step_neighbors\u001b[0;34m(pc1, pc2, params_old, lam_min, lam_max, D, sigma, max_iter, neighbors_take_each, verbose)\u001b[0m\n\u001b[1;32m    118\u001b[0m     \u001b[0mpc2_pts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpc2\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpoints\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    119\u001b[0m     \u001b[0;31m# cdist = sp.spatial.distance.cdist(pcA_pts_subset, pcB_pts)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 120\u001b[0;31m     \u001b[0mcdist\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcdist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpc1_pts_subset\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpc2_pts\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    121\u001b[0m     \u001b[0mreg_p2l\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0margmin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcdist\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maxis\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    122\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mRuntimeError\u001b[0m: CUDA out of memory. Tried to allocate 2.28 GiB (GPU 0; 10.92 GiB total capacity; 6.17 GiB already allocated; 1.83 GiB free; 2.34 GiB cached)"
     ]
    }
   ],
   "source": [
    "d1.requires_grad = True\n",
    "lam_norm = 1e-10\n",
    "opt = optim.Adam([d1], lr=0.01, amsgrad=True)\n",
    "loss_denoise = [F.mse_loss(d1.data, gt_depth[0]).cpu().numpy()]\n",
    "print('Loss_denoise = ', loss_denoise[0])\n",
    "loss_align = []\n",
    "for i in range(100):\n",
    "    pc0_noisy = Cloud.from_tensors(d1.cuda(), K.cuda(), torch.eye(4).cuda(), ignore='max')\n",
    "    pc1_noisy = Cloud.from_tensors(d2.cuda(), K.cuda(), torch.eye(4).cuda(), ignore='max')\n",
    "    print(pc0.points.shape)\n",
    "    params = perform_gradslam(pc0_noisy, pc1_noisy, gradslam_iter=15)\n",
    "    loss_val = loss_params(params, extr[0], extr[1])\n",
    "    loss_tv = l1_depth(d1)*lam_norm\n",
    "    loss_val = loss_val + loss_tv\n",
    "    opt.zero_grad()\n",
    "    loss_val.backward()\n",
    "    opt.step()\n",
    "    loss_align.append(loss_val.data.cpu().numpy())\n",
    "    loss_denoise.append(F.mse_loss(d1.data, gt_depth[0]).cpu().numpy())\n",
    "    print('Loss_align = ', loss_align[-1])\n",
    "    print('Loss_TV = ', loss_tv.data.cpu().numpy())\n",
    "    print('Loss_denoise = ', loss_denoise[-1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Without TV regularization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 415,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'Iteration number')"
      ]
     },
     "execution_count": 415,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(20, 5))\n",
    "ax[0].plot(loss_align)\n",
    "ax[0].set_title('Alignment loss')\n",
    "ax[0].set_xlabel('Iteration number')\n",
    "\n",
    "ax[1].plot(loss_denoise)\n",
    "ax[1].set_title('Denoising loss')\n",
    "ax[1].set_xlabel('Iteration number')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## With TV regularization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'Iteration number')"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(nrows=1, ncols=2, figsize=(20, 5))\n",
    "ax[0].plot(loss_align)\n",
    "ax[0].set_title('Alignment loss')\n",
    "ax[0].set_xlabel('Iteration number')\n",
    "\n",
    "ax[1].plot(loss_denoise)\n",
    "ax[1].set_title('Denoising loss')\n",
    "ax[1].set_xlabel('Iteration number')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}