after balance dataset

Pesudo_label.ipynb

@@ -0,0 +1,120 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 37,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import mmcv\n",
+    "import cv2\n",
+    "import numpy as np\n",
+    "from tqdm import tqdm_notebook as tqdm"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "out_file = \"./result/test_postnms.pkl\"\n",
+    "ann = mmcv.load(out_file)\n",
+    "CLASSES = ['tennis-court', 'container-crane', 'storage-tank', 'baseball-diamond', 'plane', 'ground-track-field', 'helicopter', 'airport', 'harbor', 'ship', 'large-vehicle', 'swimming-pool', 'soccer-ball-field', 'roundabout', 'basketball-court', 'bridge', 'small-vehicle', 'helipad']"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 42,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def generate_single_label(info, img_name, outdir, score_thresh=0.5):\n",
+    "    bboxes = info['bbox']\n",
+    "    polys = info['vis']\n",
+    "    fp = open(outdir + \"{}.txt\".format(img_name), \"w\")\n",
+    "    fp.write(\"xx\\nxx\\n\")\n",
+    "    for i in range(len(CLASSES)):\n",
+    "        class_name = CLASSES[i]\n",
+    "        bbox = bboxes[i]\n",
+    "        poly = polys[i]\n",
+    "        if len(bbox) == 0:\n",
+    "            continue\n",
+    "        scores = bbox[:,4]\n",
+    "        for p, score in zip(poly, scores):\n",
+    "            if(score<0.5):\n",
+    "                continue\n",
+    "            rect = np.array(p).flatten()\n",
+    "            rect = [str(x) for x in rect]\n",
+    "            loc = \" \".join(rect)\n",
+    "            out = loc + \" \" + class_name + \" 0\\n\"\n",
+    "            fp.write(out)\n",
+    "    fp.close()\n",
+    "\n",
+    "def generate_labels(ann, outdir, score_thresh=0.5):\n",
+    "    img_names = ann.keys()\n",
+    "    for img_name in tqdm(img_names):\n",
+    "         generate_single_label(ann[img_name], img_name, outdir, score_thresh)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 43,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9a6eff1102494dd98e264c229d098380",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "HBox(children=(IntProgress(value=0, max=780), HTML(value='')))"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "generate_labels(ann, \"./result/pesudo/labelTxt/\", 0.5)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "mm",
+   "language": "python",
+   "name": "mmdet"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.6.8"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}