{
 "cells": [
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T07:38:11.097948Z",
     "start_time": "2024-08-28T07:38:10.051274Z"
    }
   },
   "cell_type": "code",
   "source": [
    "import json\n",
    "\n",
    "from ultralytics import YOLO\n",
    "import os\n",
    "\n",
    "from export_prediction_to_label_studio import export_one_yolo_output_to_label_studio_rectangles"
   ],
   "id": "688b209f5c7b833e",
   "outputs": [],
   "execution_count": 1
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T07:38:24.127255Z",
     "start_time": "2024-08-28T07:38:23.889160Z"
    }
   },
   "cell_type": "code",
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ],
   "id": "bcbe8835eff23931",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "execution_count": 3
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T07:38:24.905478Z",
     "start_time": "2024-08-28T07:38:24.890872Z"
    }
   },
   "cell_type": "code",
   "source": "yolo_model_loaded = YOLO(\"shared/yolov8n-fashionpedia-1.torchscript\", task='detect')",
   "id": "f427e4ba2cff784",
   "outputs": [],
   "execution_count": 4
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T07:38:31.339742Z",
     "start_time": "2024-08-28T07:38:31.325405Z"
    }
   },
   "cell_type": "code",
   "source": "root_dataset_folder = '/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k'",
   "id": "f6b614baa08e709",
   "outputs": [],
   "execution_count": 5
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T07:42:03.489627Z",
     "start_time": "2024-08-28T07:39:47.352446Z"
    }
   },
   "cell_type": "code",
   "source": [
    "for root, dirs, files in os.walk(os.path.join(root_dataset_folder, 'images')):\n",
    "    print(root)\n",
    "    for a_file in files:\n",
    "        if a_file.endswith('.jpg') or a_file.endswith('.jpeg') or a_file.endswith('.png'):\n",
    "            result = yolo_model_loaded([os.path.join(root, a_file)], verbose=False)[0]\n",
    "            current_folder = root.split('/')[-1]\n",
    "            export_one_yolo_output_to_label_studio_rectangles(\n",
    "                result,\n",
    "                'yolov8n-fashionpedia-1-torchscript-c98616b267958c24874073ce091a977ce25489f5',\n",
    "                f'/data/local-files/?d=clothes_pattern/images/{current_folder}/{a_file}',\n",
    "                root.replace('images/', 'pre-annotations/'),\n",
    "                label_studio_from_name='clothes_pattern'\n",
    "            )"
   ],
   "id": "66db1d64f1f26295",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/plisse_drape_fronce\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/jersey cotele\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/non_classe\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/rayure\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/motif repete\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/liberty\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/motif place\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/pied de poule\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/animal\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/pois\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/pois_monochrome\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/carreau\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/paillette\n",
      "/home/louis/vetement_ai/ai_that_look/pattern_detection/dataset/dataset-clothes-8-pattern-v0-labeled-15k/images/floral\n"
     ]
    }
   ],
   "execution_count": 8
  },
  {
   "metadata": {},
   "cell_type": "code",
   "outputs": [],
   "execution_count": null,
   "source": "",
   "id": "513093f0ca4ac231"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-07T14:10:47.957914Z",
     "start_time": "2024-08-07T14:10:47.920287Z"
    }
   },
   "cell_type": "code",
   "source": [
    "kids_output = [\n",
    "    \"body\",\n",
    "    \"chemise\",\n",
    "    \"combinaison\",\n",
    "    \"gigoteuse\",\n",
    "    \"gilet\",\n",
    "    \"jupe\",\n",
    "    \"pantalon\",\n",
    "    \"pyjama\",\n",
    "    \"robe\",\n",
    "    \"tshirt\"\n",
    "]\n",
    "\n",
    "attributes_to_transform_default = {\n",
    "    \"shirt, blouse\": \"chemise\",\n",
    "    \"top, t-shirt, sweatshirt\": \"tshirt\",\n",
    "    \"sweater\": \"gilet\",\n",
    "    \"cardigan\": \"gilet\",\n",
    "    \"jacket\": \"gilet\",\n",
    "    \"vest\": \"gilet\",\n",
    "    \"pants\": \"pantalon\",\n",
    "    \"shorts\": \"pantalon\",\n",
    "    \"skirt\": \"jupe\",\n",
    "    \"coat\": \"gilet\",\n",
    "    \"dress\": \"robe\",\n",
    "    \"jumpsuit\": \"combinaison\",\n",
    "}\n",
    "attributes_to_keep = {\n",
    "    \"sleeve\": \"manche\",\n",
    "    \"pocket\": \"poche\",\n",
    "    \"buckle\": \"zip\",\n",
    "    \"collar\": \"col\",\n",
    "    \"lapel\": \"revers\",\n",
    "    \"hood\": \"capuche\",\n",
    "}"
   ],
   "id": "aa14e86f94358cb1",
   "outputs": [],
   "execution_count": 27
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-07T14:53:59.250707Z",
     "start_time": "2024-08-07T14:53:57.334490Z"
    }
   },
   "cell_type": "code",
   "source": [
    "attributes_to_transform = {k: 'pyjama' for k in attributes_to_transform_default.keys()}\n",
    "\n",
    "attributes_to_label = {\n",
    "    **attributes_to_transform,\n",
    "    **attributes_to_keep,\n",
    "}\n",
    "\n",
    "assert len(set(attributes_to_transform.keys()).intersection(set(attributes_to_transform.values()))) == 0\n",
    "\n",
    "score_threshold = 0.5\n",
    "for root, dirs, files in os.walk(os.path.join(root_dataset_folder, 'pre-annotations', 'pyjama')):\n",
    "    for a_file in files:\n",
    "        if a_file.endswith('.json'):\n",
    "            filepath = os.path.join(root, a_file)\n",
    "            with open(filepath, 'r') as f:\n",
    "                file_content_json = json.load(f)\n",
    "\n",
    "            predictions = file_content_json['predictions']\n",
    "            annotations = predictions[0]['result']\n",
    "            annotations = [p for p in annotations if p['score'] >= score_threshold]\n",
    "\n",
    "            updated_annotations = []\n",
    "            for an_annotation in annotations:\n",
    "                rectangle_labels = an_annotation['value']['rectanglelabels']\n",
    "\n",
    "                updated_labels = []\n",
    "                for a_label in rectangle_labels:\n",
    "                    if a_label in attributes_to_label.keys():\n",
    "                        updated_labels.append(attributes_to_label[a_label])\n",
    "                    elif a_label in attributes_to_label.values():\n",
    "                        # to make it idempotent\n",
    "                        updated_labels.append(a_label)\n",
    "\n",
    "                if len(updated_labels) > 0:\n",
    "                    an_annotation['value']['rectanglelabels'] = updated_labels\n",
    "                    updated_annotations.append(an_annotation)\n",
    "\n",
    "            with open(filepath, 'w') as f:\n",
    "                json.dump({\n",
    "                    **file_content_json,\n",
    "                    'predictions': [\n",
    "                        {\n",
    "                            **predictions[0],\n",
    "                            \"result\": updated_annotations\n",
    "                        }\n",
    "                    ]\n",
    "                }, f)"
   ],
   "id": "4df42b4df9ca7f17",
   "outputs": [],
   "execution_count": 43
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-28T07:44:23.257797Z",
     "start_time": "2024-08-28T07:44:22.988904Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# largest surface to specific label\n",
    "score_threshold = 0.5\n",
    "folder = 'rayure'\n",
    "for root, dirs, files in os.walk(os.path.join(root_dataset_folder, 'pre-annotations', folder)):\n",
    "    for a_file in files:\n",
    "        if a_file.endswith('.json'):\n",
    "            filepath = os.path.join(root, a_file)\n",
    "            with open(filepath, 'r') as f:\n",
    "                file_content_json = json.load(f)\n",
    "\n",
    "            predictions = file_content_json['predictions']\n",
    "            annotations = predictions[0]['result']\n",
    "            annotations = list(filter(lambda x: x[\"score\"] > 0.5, annotations))\n",
    "            if len(annotations) > 0:               \n",
    "                only_annotation = max(annotations, key=lambda p: p['value']['width'] * p['value']['height'])\n",
    "                only_annotation['value']['rectanglelabels'] = [folder]\n",
    "                result = [only_annotation]\n",
    "            else:\n",
    "                result = []\n",
    "\n",
    "            with open(filepath, 'w') as f:\n",
    "                json.dump({\n",
    "                    **file_content_json,\n",
    "                    'predictions': [\n",
    "                        {\n",
    "                            **predictions[0],\n",
    "                            \"result\": result\n",
    "                        }\n",
    "                    ]\n",
    "                }, f)"
   ],
   "id": "bba8c29714c2ab7b",
   "outputs": [],
   "execution_count": 21
  },
  {
   "metadata": {},
   "cell_type": "code",
   "outputs": [],
   "execution_count": null,
   "source": "",
   "id": "262fe6ef762f57d1"
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}