reward_checking.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "f2533fa3-ae41-4a90-9c4c-85bd087ac8d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "ea48a05c-c297-4693-9573-32dd0454b232",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import h5py\n",
    "import vigra\n",
    "import numpy as np\n",
    "import elf.segmentation.features as feats\n",
    "from skimage import draw\n",
    "from skimage.filters import gaussian\n",
    "from skimage.transform import rotate\n",
    "from skimage.segmentation import relabel_sequential\n",
    "from affogato.segmentation import compute_mws_segmentation\n",
    "from elf.segmentation.features import project_node_labels_to_pixels\n",
    "from elf.segmentation.multicut import multicut_decomposition\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import cm\n",
    "import napari\n",
    "from elf.segmentation import compute_rag\n",
    "import imageio.v2 as imageio\n",
    "from rag_utils import find_dense_subgraphs\n",
    "import torch\n",
    "from graphs import get_edge_indices, collate_edges\n",
    "from lines_reward import HoughLinesReward\n",
    "from circles_reward import HoughCirclesReward"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "333083c8-890a-4742-afbd-6a8785067016",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda:0\")\n",
    "torch.cuda.set_device(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "f78a6357-c96e-47a0-8f00-6a543d0aee02",
   "metadata": {},
   "outputs": [],
   "source": [
    "def rew_f(super_pixel, pred_img, s_subgraph, RewardFunction):\n",
    "    super_pixelT = torch.from_numpy(super_pixel)\n",
    "    pred_imgT = torch.from_numpy(pred_img)\n",
    "    graph = compute_rag(super_pixelT)\n",
    "    edge_ids = graph.uvIds().T\n",
    "    edge_ids = edge_ids.astype(np.int64)\n",
    "    edge_ids, fw, inv = relabel_sequential(edge_ids)\n",
    "    edge_ids = torch.from_numpy(edge_ids)\n",
    "    edge_ids = [edge_ids]\n",
    "    dir_edge_ids = [torch.cat([_edge_ids, torch.stack([_edge_ids[1], _edge_ids[0]], dim=0)], dim=1) for _edge_ids in edge_ids]\n",
    "    _subgraphs, _sep_subgraphs = find_dense_subgraphs([eids.transpose(0, 1).cpu().numpy() for eids in edge_ids], s_subgraph)\n",
    "    \n",
    "    bs = 1\n",
    "    _subgraphs = [torch.from_numpy(sg.astype(np.int64)).permute(2, 0, 1) for sg in _subgraphs]\n",
    "    _sep_subgraphs = [torch.from_numpy(sg.astype(np.int64)).permute(2, 0, 1) for sg in _sep_subgraphs]\n",
    "    _edge_ids, (n_offs, e_offs) = collate_edges(edge_ids)\n",
    "    subgraphs, sep_subgraphs = [], []\n",
    "    \n",
    "    for i in range(len(s_subgraph)):\n",
    "        subgraphs.append(torch.cat([sg + n_offs[i] for i, sg in enumerate(_subgraphs[i*bs:(i+1)*bs])], -2).flatten(-2, -1))\n",
    "        sep_subgraphs.append(torch.cat(_sep_subgraphs[i*bs:(i+1)*bs], -2).flatten(-2, -1))\n",
    "\n",
    "    SI = get_edge_indices(_edge_ids, subgraphs)\n",
    "    \n",
    "    reward_function = RewardFunction(s_subgraph)\n",
    "\n",
    "    pred_imgT = torch.unsqueeze(pred_imgT, 0)\n",
    "    super_pixelT = torch.unsqueeze(super_pixelT, 0)\n",
    "\n",
    "    reward = reward_function(prediction_segmentation = pred_imgT,\n",
    "                             gt = None, dir_edges = dir_edge_ids,\n",
    "                             superpixel_segmentation = super_pixelT,\n",
    "                             node_feats = None, actions = None,\n",
    "                             subgraph_indices = SI, sg_gt_edges = None)\n",
    "\n",
    "    return reward\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "693701d2-331a-49fa-8bae-3d6f3798c6cd",
   "metadata": {},
   "source": [
    "## Lines with Line Reward Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "713ac3c3-2bd9-4f88-9f75-cdba1d5545fc",
   "metadata": {},
   "outputs": [],
   "source": [
    "main_img = imageio.imread(\"/home/zarvandi/unsup_pix_embed/line.png\")\n",
    "main_img = torch.from_numpy(main_img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "39930d3b-9f0f-46cb-ab8a-513fdd7bab98",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f7fc4003d90>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(main_img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "059d94e9-b9b8-48e6-b78e-d370b6389567",
   "metadata": {},
   "outputs": [],
   "source": [
    "super_pixel = imageio.imread(\"/home/zarvandi/unsup_pix_embed/suppix.tif\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "128910aa-3532-4784-811e-21d4000202b1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f80287cb8b0>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(super_pixel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ca1ca82c-d317-47d2-9051-b4ee53efcd27",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_img1 = imageio.imread(\"/home/zarvandi/unsup_pix_embed/merged1.tif\")\n",
    "pred_img2 = imageio.imread(\"/home/zarvandi/unsup_pix_embed/merged2.tif\")\n",
    "pred_img3 = imageio.imread(\"/home/zarvandi/unsup_pix_embed/merged3.tif\")\n",
    "pred_img4 = imageio.imread(\"/home/zarvandi/unsup_pix_embed/merged4.tif\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ab475dee-b5cf-437c-8d20-bcb63a3f383b",
   "metadata": {},
   "outputs": [],
   "source": [
    "s_subgraph = [4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "35602d86-a13a-4c4a-b29f-d99a5389a917",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0558, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img1, s_subgraph, HoughLinesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "db6e302d-1e98-454c-944d-b7902a2ee54f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img2, s_subgraph, HoughLinesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "1f7e1b4f-05a1-4450-8115-91e35ad5d76a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img3, s_subgraph, HoughLinesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "0919e88c-9c73-429a-88d1-ae6eac46575c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0558, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img4, s_subgraph, HoughLinesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "46ecabb0-1bba-4b86-878f-ebac339b493e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# zero_img = np.zeros_like(img)\n",
    "# np.allclose(img, zero_img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "d81ac5db-29ed-43f6-b26b-ad4e6116153c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# zeroT_img = torch.zeros_like(img)\n",
    "# torch.allclose(img, zeroT_img)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3c4b81ea-842a-48e1-90f3-3efb5eb16bd5",
   "metadata": {},
   "source": [
    "## Lines with Circle Reward Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "15699cdc-02a6-46ff-9c07-71e2c390a779",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img1, s_subgraph, HoughCirclesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "aa6523f5-6f68-4e99-b605-c17dc64d3a5c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0558, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img2, s_subgraph, HoughCirclesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "6bc92774-a118-4cc0-97bf-9845fef721fa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0558, 0.0558, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img3, s_subgraph, HoughCirclesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "d36708b8-02cd-4d9e-ba52-09378b3ec02d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0558, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0542))"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel, pred_img4, s_subgraph, HoughCirclesReward)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ab62340-922a-40f8-adff-437fe58efb22",
   "metadata": {},
   "source": [
    "## Circles with Line Reward function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "11defb40-5b16-46ea-99d2-44a56d714036",
   "metadata": {},
   "outputs": [],
   "source": [
    "main_img_crl = imageio.imread(\"/home/zarvandi/unsup_pix_embed/circle.png\")\n",
    "main_img_crl = torch.from_numpy(main_img_crl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "2a84639d-2cb2-4d8a-b162-b302e2fe3543",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7fc6300b3c40>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(main_img_crl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "318d8ced-97b0-4a34-9506-79d5d64dbec4",
   "metadata": {},
   "outputs": [],
   "source": [
    "super_pixel_crl = imageio.imread(\"/home/zarvandi/unsup_pix_embed/suppix_crl.tif\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a88f8255-37cc-4c7d-ae6d-10b100c2fe1b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f739552ee50>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(super_pixel_crl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "3fca59c2-dade-4588-ac6f-64582ea9b42e",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_img1_crl = imageio.imread(\"/home/zarvandi/unsup_pix_embed/merged1_crl.tif\")\n",
    "pred_img2_crl = imageio.imread(\"/home/zarvandi/unsup_pix_embed/merged2_crl.tif\")\n",
    "pred_img3_crl = imageio.imread(\"/home/zarvandi/unsup_pix_embed/merged3_crl.tif\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "422006e9-4c8b-4fac-b8b9-4276ed461524",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0534))"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel_crl, pred_img1_crl, s_subgraph, HoughLinesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a53d50db-cd6a-4896-bdea-c5ed3672350f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0534))"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel_crl, pred_img2_crl, s_subgraph, HoughLinesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "6cf83c2a-de59-4a04-baa6-46a6d35df8a0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0534))"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel_crl, pred_img3_crl, s_subgraph, HoughLinesReward)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2135125-da28-457f-9bae-044f4c1ca293",
   "metadata": {},
   "source": [
    "## Circles with Circle Reward Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "f8e93581-3581-4738-8478-c2369033862f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0534))"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel_crl, pred_img1_crl, s_subgraph, HoughCirclesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "5d04615e-9caa-4303-9444-fdeefeb5a35b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0534))"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel_crl, pred_img2_crl, s_subgraph, HoughCirclesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "492c8d28-4457-4f85-872a-6e007145694c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([tensor([0.0538, 0.0538, 0.0538, 0.0538, 0.0538])],\n",
       " [tensor([0.0498, 0.0498, 0.0498, 0.0498, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000])],\n",
       " tensor(0.0534))"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rew_f(super_pixel_crl, pred_img3_crl, s_subgraph, HoughCirclesReward)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "17769749-0702-44a0-82f7-bc854fa54777",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}