affine_grid_gen.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "35aa88fb-7317-4843-a702-25047f19a5d5",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "f10647e3-b2c0-42c4-90c8-3ad3cadd7552",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "\n",
    "theta = torch.tensor([\n",
    "    [1.1, 1.2, 0.3],\n",
    "    [1.4, 1.5, 0.6],\n",
    "])[None, :, :]\n",
    "\n",
    "size = (1, 1, 32, 32)\n",
    "\n",
    "grid = torch.affine_grid_generator(theta, size, align_corners=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "bb13313e-7772-4455-8298-afdaef2b17b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "img = torch.zeros(1, 1, 32, 32)\n",
    "img[0, 0, 10:24, 10:24] = 1.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "0ed557e5-c347-47b8-a620-922ded0a0899",
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.nn.functional import grid_sample\n",
    "\n",
    "mode = \"bilinear\"\n",
    "out = grid_sample(img, grid, mode=mode, padding_mode=\"zeros\", align_corners=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "e87aaca2-af0c-4030-adb4-059e1c266040",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f0aa922b910>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "\n",
    "plt.figure(figsize=(12, 7))\n",
    "plt.subplot(121)\n",
    "plt.imshow(img[0, 0, ...])\n",
    "plt.subplot(122)\n",
    "plt.imshow(out[0, 0, ...])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c1fe6a79-3c14-4baa-8ee0-ec934ec6e5e6",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c4b15861-be82-407e-85d0-81ee6732941b",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}