isolated nlm benchmarking notebook

src/scripts/publication/Benchmarks_nlm.ipynb

@@ -0,0 +1,107 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "pc_name = \"BrunoMacM1Air\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "import yaml\n",
+    "\n",
+    "import numpy as np\n",
+    "from nanopyx.core.transform import NLMDenoising\n",
+    "\n",
+    "def concatenate_benchmarks(results, args_repr, previous_dict):\n",
+    "    if args_repr not in previous_dict:\n",
+    "        previous_dict[args_repr] = {}\n",
+    "        \n",
+    "    for t2run,name,_ in results:\n",
+    "        if name in previous_dict[args_repr]:\n",
+    "            previous_dict[args_repr][name].append(t2run)\n",
+    "        else:\n",
+    "            previous_dict[args_repr][name] = [t2run]\n",
+    "    \n",
+    "    return previous_dict\n",
+    "\n",
+    "\n",
+    "denoising = NLMDenoising(verbose=False,testing=True)\n",
+    "denoising._run_types.pop('threaded_dynamic')\n",
+    "denoising._run_types.pop('threaded_static')\n",
+    "denoising._run_types.pop('threaded_guided')\n",
+    "denoising._run_types.pop('python')\n",
+    "\n",
+    "denoising_results = {}\n",
+    "\n",
+    "n_repeats = 5"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "nlm_image_1 = np.random.random((1, 1000, 1000)).astype(np.float32)\n",
+    "nlm_image_2 = np.random.random((1, 500, 500)).astype(np.float32)\n",
+    "nlm_image_3 = np.random.random((1, 200, 200)).astype(np.float32)\n",
+    "\n",
+    "args_1 = [nlm_image_1]\n",
+    "kwargs_1 = {'patch_size': 50, 'patch_distance': 50, 'h': 0.1, 'sigma': 1.0}\n",
+    "args_repr_1,_ = conv._get_args_repr_score(*args_1, **kwargs_1)\n",
+    "\n",
+    "args_2 = [nlm_image_2]\n",
+    "kwargs_2 = {'patch_size': 5, 'patch_distance': 100, 'h': 0.1, 'sigma': 1.0}\n",
+    "args_repr_2,_ = conv._get_args_repr_score(*args_2, **kwargs_2)\n",
+    "\n",
+    "args_3 = [nlm_image_3]\n",
+    "kwargs_3 = {'patch_size': 5, 'patch_distance': 10, 'h': 0.1, 'sigma': 1.0}\n",
+    "args_repr_3,_ = conv._get_args_repr_score(*args_3, **kwargs_3)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "for i in range(n_repeats): \n",
+    "    results = denoising.benchmark(*args_1, **kwargs_1)\n",
+    "    denoising_results = concatenate_benchmarks(results, args_repr_1, denoising_results)\n",
+    "    results = denoising.benchmark(*args_2, **kwargs_2)\n",
+    "    denoising_results = concatenate_benchmarks(results, args_repr_2, denoising_results)\n",
+    "    results = denoising.benchmark(*args_3, **kwargs_3)\n",
+    "    denoising_results = concatenate_benchmarks(results, args_repr_3, denoising_results)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "if not os.path.exists(\"Benchmark_results\"):\n",
+    "    os.mkdir(\"Benchmark_results\")\n",
+    "\n",
+    "with open(f\"Benchmark_results/{pc_name}_Results_denoising.yaml\", 'w') as f:\n",
+    "    yaml.dump(denoising_results, f)"
+   ]
+  }
+ ],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}