init-GrandPrix+dataset_v1-15kps

Grand_Prix/Inference/predict.ipynb

@@ -0,0 +1,54 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "120b4212",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!python -m openpifpaf.predict test\\images\\test.png \\\n",
+    "--checkpoint=model/mobilenetv2-GrandPrix_v1_15.pkl.№1.epoch510 --image-output \\\n",
+    "--keypoint-threshold-rel=0.01 --cif-th=0.01 --caf-th=0.01 --seed-threshold=0.01 \\\n",
+    "--keypoint-threshold=0.01 --tr-minimum-threshold=0.01 --tr-single-pose-threshold=0.01 \\\n",
+    "--tr-multi-pose-threshold=0.01 --instance-threshold=0.1 --line-width=2 --json-output"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "984d60d5",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# !python -m openpifpaf.video --source test\\videos\\test.mp4 \\\n",
+    "# --checkpoint=model/mobilenetv2-GrandPrix_v1_15.pkl.№1.epoch510 --video-output --json-output \\\n",
+    "# --keypoint-threshold-rel=0.01 --cif-th=0.01 --caf-th=0.01 --seed-threshold=0.01 \\\n",
+    "# --keypoint-threshold=0.01 --tr-minimum-threshold=0.01 --tr-single-pose-threshold=0.01 \\\n",
+    "# --tr-multi-pose-threshold=0.01 --instance-threshold=0.1 --line-width=4 --video-fps=25 \\\n",
+    "# --scale=1 --loader-workers=3"
+   ]
+  }
+ ],
+ "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
+}

Grand_Prix/Inference/test/images/test.png.predictions.json

@@ -0,0 +1 @@
+[{"keypoints": [838.75, 615.73, 0.88, 909.25, 613.23, 0.96, 1016.17, 620.18, 0.97, 1061.86, 628.27, 0.97, 1009.92, 566.12, 0.97, 1058.38, 501.94, 0.97, 1009.72, 476.08, 0.97, 936.14, 503.76, 0.99, 887.17, 524.44, 1.0, 800.59, 530.07, 0.99, 692.88, 533.28, 0.93, 649.33, 581.98, 0.97, 703.06, 585.64, 0.96, 912.56, 581.72, 0.96, 921.69, 479.73, 0.87], "bbox": [639.74, 465.77, 433.49, 173.86], "score": 0.957, "category_id": 1}]

Grand_Prix/Supervisely/Bbox from Pose.ipynb

@@ -0,0 +1,227 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "07bdc0da",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "import json\n",
+    "import numpy as np\n",
+    "from tqdm.notebook import tqdm\n",
+    "path_dataset = r'D:\\GrandPrix\\Grand-Prix'"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "ca890aa0",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def find_image(id):\n",
+    "    for row in coco['images']:\n",
+    "        if row['id'] == id:\n",
+    "            return row"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "b8135031",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def ltrb_from_cloud(cloud_2d, imgSize, expansion=0.1):\n",
+    "    height, width = imgSize\n",
+    "    ltrb = np.round((cloud_2d[:, 0].min(), cloud_2d[:, 1].min(),\n",
+    "                     cloud_2d[:, 0].max(), cloud_2d[:, 1].max())).astype(int)\n",
+    "    \n",
+    "    if expansion > 0:\n",
+    "        dx = np.round((ltrb[2]-ltrb[0])*expansion/2)\n",
+    "        dy = np.round((ltrb[3]-ltrb[1])*expansion/2)\n",
+    "        ltrb += np.array([-dx, -2*dy, dx, dy], dtype=int)\n",
+    "        \n",
+    "    ltrb[[0,2]] = np.clip(ltrb[[0,2]], 0, width)\n",
+    "    ltrb[[1,3]] = np.clip(ltrb[[1,3]], 0, height)\n",
+    "    \n",
+    "    return ltrb\n",
+    "\n",
+    "def ltrb2ltwh(ltrb):\n",
+    "    return np.array([ltrb[0], ltrb[1], ltrb[2]-ltrb[0], ltrb[3]-ltrb[1]], dtype=int)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "79762c08",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open(os.path.join(path_dataset, 'GrandPrix_COCO.json'), 'r') as file:\n",
+    "    coco = json.load(file)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a1cfe0e2",
+   "metadata": {},
+   "source": [
+    "### Bboxes from pose"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "f9440b54",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "758138a872a9485885387a5ffe5a9696",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/148 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "for obj in tqdm(coco['annotations']):\n",
+    "    keypoints_2d = np.array(obj['keypoints']).reshape((-1, 3))[:, :2]\n",
+    "    id, width, height, file_name = find_image(obj['image_id']).values()\n",
+    "    \n",
+    "    bbox_ltrb = ltrb_from_cloud(keypoints_2d, (height, width), 0.4)\n",
+    "    bbox_ltwh = ltrb2ltwh(bbox_ltrb).tolist()\n",
+    "    obj['bbox'] = bbox_ltwh\n",
+    "    obj['area'] = bbox_ltwh[2]*bbox_ltwh[3]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "dfcb675b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open(os.path.join(path_dataset, 'GrandPrix_COCO.json'), 'w') as file:\n",
+    "    json.dump(coco, file)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e8ff698d",
+   "metadata": {},
+   "source": [
+    "### Split COCO json to train/val"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "3aec7a2b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from sklearn.model_selection import train_test_split\n",
+    "import copy\n",
+    "\n",
+    "def split_coco_json(coco, test_size=0.2, random_state=0):\n",
+    "    \n",
+    "    train_idx, test_idx = train_test_split([i['id'] for i in coco['images']],\n",
+    "                                           test_size=test_size, random_state=random_state)\n",
+    "\n",
+    "\n",
+    "    train = copy.deepcopy(coco)\n",
+    "    test = copy.deepcopy(coco)\n",
+    "\n",
+    "    test['images'] = [x for x in coco['images'] if x['id'] in test_idx]\n",
+    "    train['images'] = [x for x in coco['images'] if x['id'] in train_idx]\n",
+    "\n",
+    "    test['annotations'] = [x for x in coco['annotations'] if x['image_id'] in test_idx]\n",
+    "    train['annotations'] = [x for x in coco['annotations'] if x['image_id'] in train_idx]\n",
+    "    return train, test"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "1cdd1a3c",
+   "metadata": {},
+   "source": [
+    "### Create new splited dataset"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "aafd54fb",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train, test = split_coco_json(coco, 0.1, random_state=77)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "f6af98fc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test_path_images = [os.path.join(path_dataset, 'images', 'img', x['file_name']) for x in test['images']]\n",
+    "train_path_images = [os.path.join(path_dataset, 'images', 'img', x['file_name']) for x in train['images']]\n",
+    "\n",
+    "import shutil\n",
+    "path_new_dataset = r'C:\\Users\\Kir\\Jupiter\\Drivecast\\place6D_Nurburgring\\Grand_Prix\\Training\\GrandPrix_dataset_v1_15'\n",
+    "\n",
+    "path_train_img = os.path.join(path_new_dataset, 'images', 'train')\n",
+    "path_test_img = os.path.join(path_new_dataset, 'images', 'val')\n",
+    "path_ann = os.path.join(path_new_dataset, 'annotations')\n",
+    "\n",
+    "os.makedirs(path_train_img, exist_ok=True)\n",
+    "os.makedirs(path_test_img, exist_ok=True)\n",
+    "os.makedirs(path_ann, exist_ok=True)\n",
+    "\n",
+    "with open(os.path.join(path_ann, 'train.json'), 'w') as file:\n",
+    "    json.dump(train, file)\n",
+    "    \n",
+    "with open(os.path.join(path_ann, 'val.json'), 'w') as file:\n",
+    "    json.dump(test, file)\n",
+    "\n",
+    "for path in train_path_images:\n",
+    "    shutil.copy(path, path_train_img)\n",
+    "\n",
+    "for path in test_path_images:\n",
+    "    shutil.copy(path, path_test_img)"
+   ]
+  }
+ ],
+ "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
+}

Grand_Prix/Supervisely/GrandPrix_create_3d_15.ipynb

@@ -0,0 +1,83 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "9b13894c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import json\n",
+    "import pandas as pd\n",
+    "import numpy as np\n",
+    "import cv2 \n",
+    "import plotly.express as px\n",
+    "import plotly.graph_objects as go\n",
+    "# from utils import *\n",
+    "\n",
+    "def roration(x):\n",
+    "    x = np.deg2rad(x)\n",
+    "    return np.array([[np.cos(x), -np.sin(x)],\n",
+    "                    [np.sin(x), np.cos(x)]])\n",
+    "\n",
+    "arrow = np.float32([[0,1,0],\n",
+    "                  [0,-1,0]])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 178,
+   "id": "39604368",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "img = cv2.imread('Grand_Prix_GPS-double.png')\n",
+    "df = pd.read_csv('raw_keypoints.csv',index_col='index')\n",
+    "df.loc[[1,2], ['x','y']] -=3\n",
+    "df_raw = df.copy()\n",
+    "center = (1164, 703)\n",
+    "df[['x', 'y']]-= center\n",
+    "df = df.astype('float32')\n",
+    "df['z'] = 4.2*np.float32([0, 0, 0, 0, 3, 6.33, 7.33, 3.33, 2.33, 2.33, 2.33, 2.5, 2.5, 1.2,1.5])\n",
+    "df[['x', 'y']] /= 8.4\n",
+    "df[['x', 'y']] = df.values[:, :2]@roration(-135) # поворот x и y в плоскости земли\n",
+    "# df.to_csv('GrandPrix_15kps.csv')\n",
+    "\n",
+    "# rvec, tvec, camMatrx, dist = fit(img.shape[:2], df_raw, df)\n",
+    "# imgpts = cv.projectPoints(5*arrow, rvec, tvec, camMatrx, dist)[0][:, 0].astype(int)\n",
+    "# for i, p in enumerate(imgpts):\n",
+    "#     cv2.circle(img, p, 3*(i+1), color=(200,25,200), thickness=-1)\n",
+    "# px.imshow(img)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "82d1f0f1",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "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
+}

Grand_Prix/Supervisely/ToCOCO.ipynb

@@ -0,0 +1,228 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "283f6e9c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "import numpy as np\n",
+    "import pandas as pd\n",
+    "import glob\n",
+    "import json\n",
+    "from tqdm.notebook import tqdm\n",
+    "\n",
+    "path_dataset = r'D:\\GrandPrix\\Grand-Prix'\n",
+    "with open(os.path.join(path_dataset, 'meta.json'), 'r') as j:\n",
+    "    meta = json.load(j)\n",
+    "\n",
+    "imgs = glob.glob(path_dataset + '\\\\images\\\\img\\\\*', recursive=True)\n",
+    "anns = glob.glob(path_dataset + '\\\\images\\\\ann\\\\*', recursive=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "7af948d4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def label2hash(meta_json, last):\n",
+    "    for clss in meta_json['classes']:\n",
+    "        if clss['title'] == last['classTitle']:\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"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "7ea0771e",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def annotations(meta_json, obj):\n",
+    "    nodes = obj['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, obj)[str(i)]]['loc']\n",
+    "\n",
+    "    keypoints_2d['v'] = 2\n",
+    "    keypoints_2d[(keypoints_2d.x<3)&(keypoints_2d.y<3)] = 0\n",
+    "    keypoints_2d = keypoints_2d.astype(float).round().astype(int)\n",
+    "    return keypoints_2d"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "26b6abf4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def ann_json(keypoints, img_id, obj):\n",
+    "    \n",
+    "    annotation = {\n",
+    "        \"id\": obj['id'],\n",
+    "        \"segmentation\": [],\n",
+    "        \"num_keypoints\": len(keypoints),\n",
+    "        \"area\": 0,\n",
+    "        \"iscrowd\": 0,\n",
+    "        \"image_id\": img_id,\n",
+    "        \"bbox\": [],\n",
+    "        \"category_id\": 1,\n",
+    "        \"keypoints\": keypoints.values.flatten().tolist()}\n",
+    "\n",
+    "    return annotation\n",
+    "\n",
+    "def img_json(ann, name, id):\n",
+    "    height, width =  ann['size'].values()\n",
+    "    image = {\n",
+    "        \"id\": id,\n",
+    "        \"width\": width,\n",
+    "        \"height\": height,\n",
+    "        \"file_name\": name,\n",
+    "    }\n",
+    "    return image"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "7c4d21a7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def skeleton_supervisely(meta):\n",
+    "    meta_nodes = meta['classes'][0]['geometry_config']['nodes']\n",
+    "    label2hash = {}\n",
+    "    for name in meta_nodes:\n",
+    "        label2hash[meta_nodes[name]['label']] = name\n",
+    "\n",
+    "    skeleton_supervisely = []\n",
+    "    for edge in meta['classes'][0]['geometry_config']['edges']:\n",
+    "        skeleton_supervisely.append([meta_nodes[edge['src']]['label'], meta_nodes[edge['dst']]['label']])\n",
+    "    return skeleton_supervisely"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "155be42d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def ann_img_list(anns, imgs, meta):\n",
+    "    annotations_list = []\n",
+    "    image_list = []\n",
+    "    for i in tqdm(range(len(anns))):\n",
+    "\n",
+    "        with open(anns[i], 'r') as j:\n",
+    "            ann = json.load(j)\n",
+    "        \n",
+    "        image_name = os.path.basename(anns[i])[:-5]\n",
+    "        image = img_json(ann, image_name, i)\n",
+    "        image_list.append(image)\n",
+    "\n",
+    "        for obj in ann['objects']:\n",
+    "            keypoints = annotations(meta, obj)\n",
+    "            annotations_list.append(ann_json(keypoints, i, obj))\n",
+    "    return image_list, annotations_list, meta"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "6593f677",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def COCO(image_list, annotations_list, meta):\n",
+    "    num_kpts = len(meta['classes'][0]['geometry_config']['nodes'])\n",
+    "    coco = {\n",
+    "\n",
+    "        \"info\": {\n",
+    "            \"description\": \"karusel Dataset\", \"version\": \"1.0\", \"keypoints\":num_kpts\n",
+    "        },\n",
+    "\n",
+    "        \"categories\": [\n",
+    "            {\n",
+    "                \"supercategory\": \"NurburgRing\",\n",
+    "                \"id\": 1,\n",
+    "                \"name\": \"GrandPrix\",\n",
+    "                \"keypoints\": list(range(num_kpts)),\n",
+    "                \"skeleton\": skeleton_supervisely(meta)\n",
+    "            }\n",
+    "        ]\n",
+    "    }\n",
+    "\n",
+    "    coco['images'] = image_list\n",
+    "    coco['annotations'] = annotations_list\n",
+    "    return coco"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "38880d13",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d7d6cab432b5425894ec499e3216f3a3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/148 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "coco_json = COCO(*ann_img_list(anns, imgs, meta))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "314565c2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open(os.path.join(path_dataset, 'GrandPrix_COCO.json'), 'w') as file:\n",
+    "    json.dump(coco_json, file)"
+   ]
+  }
+ ],
+ "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
+}

Grand_Prix/Supervisely/raw_keypoints.csv

@@ -0,0 +1,16 @@
+index,x,y
+1,1213,749
+2,1123,661
+3,1093,473
+4,1127,364
+5,1100,472
+6,1129,364
+7,1094,474
+8,734,841
+9,804,924
+10,918,1035
+11,1066,1183
+12,1500,960
+13,1468,868
+14,1064,693
+15,356,1107