#42 Add FIW with deepface analysis

vitalwarley · Mar 10, 2024 · c15f580 · c15f580
1 parent 238ddac
commit c15f580
Show file tree

Hide file tree

Showing 4 changed files with 26,178 additions and 0 deletions.
diff --git a/notebooks/apply deepface.ipynb b/notebooks/apply deepface.ipynb
@@ -0,0 +1,165 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from deepface import DeepFace\n",
+    "import pandas as pd\n",
+    "import os\n",
+    "\n",
+    "\n",
+    "os.chdir('../../rfiw2021')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df = pd.read_csv('Track1/sample0/val.txt', sep=\" \", header=None)\n",
+    "df.columns = ['index', 'face1_path', 'face2_path', 'kinship', 'label']\n",
+    "df.drop('index', axis=1, inplace=True) "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 112,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from tqdm import tqdm\n",
+    "\n",
+    "def analyze_batch(df, b, name):\n",
+    "    base_file = 'Track1/'\n",
+    "    \n",
+    "    age = []\n",
+    "    gender = []\n",
+    "    race = []\n",
+    "\n",
+    "    for i, data in tqdm(df.iterrows(), total=len(df), desc=f\"Processing batch {b}\"):\n",
+    "\n",
+    "        obj = DeepFace.analyze(img_path = base_file+data['face'], enforce_detection=False, silent=True,\n",
+    "            actions = ['age', 'gender', 'race']\n",
+    "        )[0]\n",
+    "\n",
+    "        age.append(obj['age'])\n",
+    "        gender.append(obj['dominant_gender'])\n",
+    "        race.append(obj['dominant_race'])\n",
+    "\n",
+    "\n",
+    "    df['age'] = age\n",
+    "    df['gender'] = gender\n",
+    "    df['race'] = race\n",
+    "\n",
+    "    df.to_csv(f'fiw/{name}-{b}.csv', index=False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 115,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Processing batch 0: 100%|██████████| 128/128 [00:50<00:00,  2.55it/s]\n",
+      "Processing batch 1: 100%|██████████| 128/128 [00:49<00:00,  2.58it/s]\n",
+      "Processing batch 2: 100%|██████████| 128/128 [00:51<00:00,  2.49it/s]\n",
+      "Processing batch 3: 100%|██████████| 128/128 [00:51<00:00,  2.48it/s]\n",
+      "Processing batch 4: 100%|██████████| 128/128 [00:50<00:00,  2.52it/s]\n",
+      "Processing batch 5: 100%|██████████| 128/128 [00:48<00:00,  2.63it/s]\n",
+      "Processing batch 6: 100%|██████████| 128/128 [00:51<00:00,  2.46it/s]\n",
+      "Processing batch 7: 100%|██████████| 128/128 [00:51<00:00,  2.49it/s]\n",
+      "Processing batch 8: 100%|██████████| 128/128 [00:48<00:00,  2.62it/s]\n",
+      "Processing batch 9: 100%|██████████| 128/128 [00:47<00:00,  2.71it/s]\n",
+      "Processing batch 10: 100%|██████████| 128/128 [00:49<00:00,  2.58it/s]\n",
+      "Processing batch 11: 100%|██████████| 128/128 [00:49<00:00,  2.61it/s]\n",
+      "Processing batch 12: 100%|██████████| 128/128 [00:49<00:00,  2.59it/s]\n",
+      "Processing batch 13: 100%|██████████| 128/128 [00:48<00:00,  2.63it/s]\n",
+      "Processing batch 14: 100%|██████████| 128/128 [00:51<00:00,  2.47it/s]\n",
+      "Processing batch 15: 100%|██████████| 128/128 [00:49<00:00,  2.59it/s]\n",
+      "Processing batch 16: 100%|██████████| 128/128 [00:48<00:00,  2.64it/s]\n",
+      "Processing batch 17: 100%|██████████| 128/128 [00:47<00:00,  2.69it/s]\n",
+      "Processing batch 18: 100%|██████████| 128/128 [00:50<00:00,  2.55it/s]\n",
+      "Processing batch 19: 100%|██████████| 128/128 [00:51<00:00,  2.48it/s]\n",
+      "Processing batch 20: 100%|██████████| 128/128 [00:49<00:00,  2.61it/s]\n",
+      "Processing batch 21: 100%|██████████| 128/128 [00:48<00:00,  2.66it/s]\n",
+      "Processing batch 22: 100%|██████████| 128/128 [00:50<00:00,  2.54it/s]\n",
+      "Processing batch 23: 100%|██████████| 128/128 [00:50<00:00,  2.55it/s]\n",
+      "Processing batch 24: 100%|██████████| 128/128 [00:49<00:00,  2.57it/s]\n",
+      "Processing batch 25: 100%|██████████| 128/128 [00:48<00:00,  2.66it/s]\n",
+      "Processing batch 26: 100%|██████████| 128/128 [00:51<00:00,  2.50it/s]\n",
+      "Processing batch 27: 100%|██████████| 128/128 [00:53<00:00,  2.39it/s]\n",
+      "Processing batch 28: 100%|██████████| 128/128 [00:49<00:00,  2.56it/s]\n",
+      "Processing batch 29: 100%|██████████| 128/128 [00:50<00:00,  2.55it/s]\n",
+      "Processing batch 30: 100%|██████████| 128/128 [00:49<00:00,  2.57it/s]\n",
+      "Processing batch 31: 100%|██████████| 128/128 [00:49<00:00,  2.59it/s]\n",
+      "Processing batch 32: 100%|██████████| 128/128 [00:52<00:00,  2.45it/s]\n",
+      "Processing batch 33: 100%|██████████| 128/128 [00:53<00:00,  2.38it/s]\n",
+      "Processing batch 34: 100%|██████████| 128/128 [00:54<00:00,  2.33it/s]\n",
+      "Processing batch 35: 100%|██████████| 128/128 [00:55<00:00,  2.32it/s]\n",
+      "Processing batch 36: 100%|██████████| 128/128 [00:55<00:00,  2.29it/s]\n",
+      "Processing batch 37: 100%|██████████| 128/128 [00:57<00:00,  2.21it/s]\n",
+      "Processing batch 38: 100%|██████████| 128/128 [00:59<00:00,  2.15it/s]\n",
+      "Processing batch 39: 100%|██████████| 128/128 [00:59<00:00,  2.16it/s]\n",
+      "Processing batch 40: 100%|██████████| 106/106 [00:48<00:00,  2.17it/s]\n"
+     ]
+    }
+   ],
+   "source": [
+    "batch_size = 128\n",
+    "\n",
+    "num_batches = len(df) // batch_size + 1 if len(df) % batch_size != 0 else len(df) // batch_size\n",
+    "\n",
+    "for i in range(num_batches):\n",
+    "    start_idx = i * batch_size\n",
+    "    end_idx = (i + 1) * batch_size\n",
+    "    batch_df = df.iloc[start_idx:end_idx]\n",
+    "\n",
+    "    if os.path.exists(f'fiw/test-{i}.csv'):\n",
+    "        continue\n",
+    "\n",
+    "    analyze_batch(batch_df, i, 'test') "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 117,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dfs = []\n",
+    "for i in range(40):\n",
+    "    dfs.append(pd.read_csv(f'fiw/test-{i}.csv'))\n",
+    "\n",
+    "df = pd.concat(dfs).to_csv('fiw/test_with_deepface.csv', index=False)"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "research",
+   "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": "3.9.18"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}