{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4c948dcd",
   "metadata": {},
   "outputs": [],
   "source": [
    "import supervisely_lib as sly\n",
    "import pandas as pd\n",
    "import cv2 as cv\n",
    "import os\n",
    "import numpy as np\n",
    "from utils import *\n",
    "from tqdm.notebook import tqdm\n",
    "import plotly.express as px\n",
    "import plotly.graph_objects as go\n",
    "\n",
    "keypoints_3d_24 = pd.read_csv('karussel_24kps.csv', index_col=0).astype(float)\n",
    "keypoints_3d_88 = pd.read_csv('keypoints_88/karussel_88kps.csv', index_col=0).astype(float)\n",
    "keypoints_3d_88.z = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "2ff23cb1",
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_ann(img_id):\n",
    "    ann_info = api.annotation.download(img_id)\n",
    "    return ann_info\n",
    "\n",
    "def label2hash(meta_json, last):\n",
    "    for clss in meta_json['classes']:\n",
    "        if clss['title'] == last['classTitle'].values[0]:\n",
    "            meta_nodes = clss['geometry_config']['nodes']\n",
    "            label2hash = {}\n",
    "            for name in meta_nodes:\n",
    "                label2hash[meta_nodes[name]['label']] = name\n",
    "    return label2hash\n",
    "\n",
    "def new_annotations(ann_info, new_keypoints, flag=False):\n",
    "    last = last_keypoints_on_img(ann_info)\n",
    "    nodes = ann_info[2]['objects'][last.index[0]]['nodes']\n",
    "    for i in new_keypoints.index:\n",
    "        if (not label2hash(meta_json, last)[str(i)] in nodes.keys()) and flag:\n",
    "            nodes[label2hash(meta_json, last)[str(i)]] = {'loc': [0, 0]}\n",
    "    \n",
    "        nodes[label2hash(meta_json, last)[str(i)]]['loc'] = new_keypoints.loc[i].tolist()\n",
    "    return ann_info\n",
    "\n",
    "def annotations(ann_info):\n",
    "\n",
    "    last = last_keypoints_on_img(ann_info)\n",
    "    if len(last) == 0:\n",
    "        return \n",
    "    nodes = ann_info[2]['objects'][last.index[0]]['nodes']\n",
    "    keypoints_2d = pd.DataFrame(columns=['x', 'y'])\n",
    "\n",
    "    for i in range(1, len(nodes)+1):\n",
    "        keypoints_2d.loc[i] = nodes[label2hash(meta_json, last)[str(i)]]['loc']\n",
    "\n",
    "    return keypoints_2d\n",
    "\n",
    "def last_keypoints_on_img(ann_info):\n",
    "    updated = []\n",
    "    for obj in ann_info[2]['objects']:\n",
    "        updated.append([obj['classTitle'], obj['updatedAt']])\n",
    "    updated = pd.DataFrame(updated, columns=['classTitle', 'updatedAt'])\n",
    "    updated.updatedAt = pd.to_datetime(updated.updatedAt)\n",
    "    last = updated[updated.updatedAt == updated.updatedAt.max()]\n",
    "    return last"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "e49d4952",
   "metadata": {},
   "outputs": [],
   "source": [
    "adress = 'https://app.supervise.ly/'\n",
    "token = '8N4spdUrCH0BxeQvv1nicgj7VDEsbpo1XUB7SKDk87woYPZPYkDo05o2jt1gpiCaiTaC1ePcWoYEi2Q2ehW5t9mghoQBShn5a6LUjfjhVRIXZ2ol2otf5HADq2EAszQP'\n",
    "project_id = 192775\n",
    "dataset_id = 658585\n",
    "api = sly.Api(adress, token)\n",
    "meta_json = api.project.get_meta(project_id)\n",
    "meta = sly.ProjectMeta.from_json(meta_json)\n",
    "images = pd.DataFrame(api.image.get_list(dataset_id)).sort_values('name', ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "558309c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# name = 'NLS4[00_04_07]006.png'\n",
    "# index = images.loc[images.name==name].index[0]\n",
    "\n",
    "\n",
    "# img = cv.imread(os.path.join(\n",
    "#     r'C:\\Users\\Kir\\Jupiter\\Nurburg\\OpenPifPaf\\Training\\Karusel_dataset_v2\\images\\val', name))\n",
    "# img = cv.cvtColor(img, cv.COLOR_BGR2RGB)\n",
    "\n",
    "# img_id = int(images.loc[index, 'id'])\n",
    "# img_name = images.loc[index, 'name']\n",
    "# ann_info = load_ann(img_id)\n",
    "# keypoints_24 = annotations(ann_info)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "50bf81ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "# a['objects'][0]['classTitle'] = '88kps-bot'\n",
    "# ann_info[2]['objects'] = a['objects']\n",
    "# keypoints_24 = annotations(ann_info)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f510305b",
   "metadata": {},
   "outputs": [],
   "source": [
    "rvec, tvec, camMatrx, dist = fit((1920, 1080), keypoints_24, keypoints_3d_24, focus=1)\n",
    "imgpts_24 = cv.projectPoints(keypoints_3d_24.values, rvec, tvec, camMatrx, dist)[0][:, 0].astype(int)\n",
    "imgpts_88 = cv.projectPoints(keypoints_3d_88.values, rvec, tvec, camMatrx, dist)[0][:, 0].astype(int)\n",
    "\n",
    "new_keypoints = pd.DataFrame(imgpts_88, columns=['x', 'y'], index=range(1, len(keypoints_3d_88)+1))\n",
    "new_keypoints[(new_keypoints.x<0) | (new_keypoints.y<0) \\\n",
    "              | (new_keypoints.x>=1920) | (new_keypoints.y>=1080)] *= 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "f619f420",
   "metadata": {},
   "outputs": [],
   "source": [
    "# fig = px.imshow(img)\n",
    "# fig.add_trace(go.Scatter(x=imgpts_24[:,0], y=imgpts_24[:,1],\n",
    "#                          marker=dict(color='white', size=5), mode='markers'))\n",
    "# fig.add_trace(go.Scatter(x=new_keypoints.values[:,0], y=new_keypoints.values[:,1],\n",
    "#                          marker=dict(color='blue', size=5), mode='markers'))\n",
    "# fig.add_trace(go.Scatter(x=keypoints_24.values[:,0], y=keypoints_24.values[:,1],\n",
    "#                          marker=dict(color='black', size=5), mode='markers'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "88c81371",
   "metadata": {},
   "outputs": [],
   "source": [
    "new_ann = new_annotations(ann_info, new_keypoints, flag=True)\n",
    "new_ann = sly.Annotation.from_json(new_ann.annotation, meta)\n",
    "api.annotation.upload_ann(img_id, new_ann)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45d35b7c",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7c2a287f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "d0dfe8b4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dd8d392171e14a2281945b8575b435df",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/448 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "error = []\n",
    "for index in tqdm(range(len(images))):\n",
    "    img_id = int(images.loc[index, 'id'])\n",
    "    img_name = images.loc[index, 'name']\n",
    "    ann_info = load_ann(img_id)\n",
    "    keypoints_24 = annotations(ann_info)\n",
    "    \n",
    "    if len(keypoints_24) != 24:\n",
    "        error.append(img_name)\n",
    "        continue\n",
    "    \n",
    "    rvec, tvec, camMatrx, dist = fit((1920, 1080), keypoints_24, keypoints_3d_24, focus=1)\n",
    "    imgpts_88 = cv.projectPoints(keypoints_3d_88.values, rvec, tvec, camMatrx, dist)[0][:, 0].astype(int)\n",
    "    new_keypoints = pd.DataFrame(imgpts_88, columns=['x', 'y'], index=range(1, len(keypoints_3d_88)+1))\n",
    "    new_keypoints[(new_keypoints.x<0) | (new_keypoints.y<0) \\\n",
    "              | (new_keypoints.x>=1920) | (new_keypoints.y>=1080)] *= 0\n",
    "\n",
    "    new_ann = new_annotations(ann_info, new_keypoints, flag=True)\n",
    "    new_ann = sly.Annotation.from_json(new_ann.annotation, meta)\n",
    "    api.annotation.upload_ann(img_id, new_ann)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e0e32ce4",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8b70bdcf",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0d04e0a8",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6713e0ed",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# no_ann = []\n",
    "# for index in tqdm(range(len(images))):\n",
    "#     img_id = int(images.loc[index, 'id'])\n",
    "#     img_name = images.loc[index, 'name']\n",
    "#     ann_info = load_ann(img_id)\n",
    "\n",
    "#     json_df = pd.read_json(os.path.join(folder,img_name+'.predictions.json'))\n",
    "#     if len(json_df)>0:\n",
    "#         keypoints_2d = json_df.sort_values('score').reset_index().loc[0, 'keypoints']\n",
    "#         keypoints_2d = pd.DataFrame(np.array(keypoints_2d).reshape(-1, 3),\n",
    "#                                     columns=['x','y','conf'], index=range(1, 25))\n",
    "\n",
    "#         conf_keypoints_2d = keypoints_2d[keypoints_2d['conf']>0.15]\n",
    "#         if len(conf_keypoints_2d)>5:\n",
    "#             rvec, tvec, camMatrx, dist = fit((1920, 1080), conf_keypoints_2d, keypoints_3d, focus=1)\n",
    "\n",
    "#             imgpts = cv.projectPoints(keypoints_3d.values, rvec, tvec, camMatrx, dist)[0][:, 0].astype(int)\n",
    "#             new_keypoints = pd.DataFrame(imgpts, columns=['x', 'y'], index=range(1, len(keypoints_3d)+1))\n",
    "#         else:   \n",
    "#             new_keypoints = keypoints_2d[['x', 'y']].astype(int)\n",
    "\n",
    "#     else:\n",
    "#         no_ann.append(img_name)\n",
    "\n",
    "\n",
    "#     new_ann = new_annotations(ann_info, new_keypoints)\n",
    "#     new_ann = sly.Annotation.from_json(new_ann.annotation, meta)\n",
    "#     api.annotation.upload_ann(img_id, new_ann)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3ab1e73a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4d632830",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a97bbbc3",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c8598689",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0974e494",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "18a70aa4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# \n",
    "# import glob\n",
    "# folder = r'C:\\Users\\Kir\\Downloads\\NLS\\images'\n",
    "# list_png = glob.glob(folder+'\\*.png')\n",
    "# list_json = glob.glob(folder+'\\*.json')\n",
    "\n",
    "# df = pd.DataFrame(np.array([list_png, list_json]).T, columns=['png', 'json'])\n",
    "\n",
    "\n",
    "# def funk(str):\n",
    "#     str = os.path.basename(str)\n",
    "#     return os.path.join(folder, str[4:8]+'['+str[27:35]+']'+str[41:44]+'.png')\n",
    "# funk(df.png[40])\n",
    "\n",
    "# df['png_new'] = df['png'].apply(funk) \n",
    "# df['json_new'] = df['json'].apply(funk) \n",
    "\n",
    "# for i in range(len(df)):\n",
    "#     os.rename(df['png'][i], df['png_new'][i])\n",
    "#     os.rename(df['json'][i], df['json_new'][i])"
   ]
  }
 ],
 "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
}