update supervisely utils

2022-05-18 12:01:01 +03:00
parent 49df17fb9c
commit f6af1d69a1
9 changed files with 516 additions and 0 deletions
--- a/Karussell/Supervisely/Bbox
+++ b/Karussell/Supervisely/Bbox
@@ -0,0 +1,217 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "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:\\karusel'\n",
+    "from SuperviselyKeypointsGUI.SuperviselyKeypointsGUI import *\n",
+    "\n",
+    "\n",
+    "keypoints_3d_path = r'SuperviselyKeypointsGUI\\karussel_24kps.csv'\n",
+    "keypoints_3d = pd.read_csv(keypoints_3d_path, index_col=0).astype(float)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "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": null,
+   "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": null,
+   "id": "79762c08",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open(os.path.join(path_dataset, 'karusel_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": null,
+   "id": "f9440b54",
+   "metadata": {},
+   "outputs": [],
+   "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": null,
+   "id": "dfcb675b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# with open(os.path.join(path_dataset, 'karusel_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": null,
+   "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": null,
+   "id": "aafd54fb",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train, test = split_coco_json(coco, 0.1, random_state=777)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "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\\Nurburg\\OpenPifPaf\\Training\\Karusel_dataset'\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
+}
--- a/Karussell/Supervisely/SuperviselyKeypointsGUI/Icon.ico
+++ b/Karussell/Supervisely/SuperviselyKeypointsGUI/Icon.ico
--- a/Karussell/Supervisely/SuperviselyKeypointsGUI/SuperviselyAPI_requirements.txt
+++ b/Karussell/Supervisely/SuperviselyKeypointsGUI/SuperviselyAPI_requirements.txt
@@ -0,0 +1,5 @@
+pip install wxPython==4.1.1
+pip install cv2
+pip install numpy
+pip install pandas
+pip install supervisely
--- a/Karussell/Supervisely/SuperviselyKeypointsGUI/SuperviselyAPI_start.ipynb
+++ b/Karussell/Supervisely/SuperviselyKeypointsGUI/SuperviselyAPI_start.ipynb
@@ -0,0 +1,53 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "a916d355",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from SuperviselyKeypointsGUI import *\n",
+    "\n",
+    "keypoints_3d_path = r'karussel_24kps.csv'\n",
+    "token = ''\n",
+    "dataset_id = 627375 #(images)\n",
+    "project_id = 184347 #(Nurburg-karussel)\n",
+    "keypoints_3d = pd.read_csv(keypoints_3d_path, index_col=0).astype(float)\n",
+    "sp = Start_annotation(project_id, dataset_id, token,\n",
+    "                     'local_dataset_path', keypoints_3d_path, 'point_cloud_path')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "d4037523",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "sp.start()"
+   ]
+  }
+ ],
+ "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
+}
--- a/Karussell/Supervisely/SuperviselyKeypointsGUI/SuperviselyKeypointsGUI.py
+++ b/Karussell/Supervisely/SuperviselyKeypointsGUI/SuperviselyKeypointsGUI.py
@@ -0,0 +1,250 @@
+import supervisely_lib as sly
+import pandas as pd
+import cv2 as cv
+import os
+# from PIL import Image
+import numpy as np
+# import open3d as o3d
+import wx
+import wx.xrc
+
+
+def last_keypoints_on_img(ann_info):
+    updated = []
+    for obj in ann_info[2]['objects']:
+        updated.append([obj['classTitle'], obj['updatedAt']])
+    updated = pd.DataFrame(updated, columns=['classTitle', 'updatedAt'])
+    updated.updatedAt = pd.to_datetime(updated.updatedAt)
+    last = updated[updated.updatedAt == updated.updatedAt.max()]
+    return last
+
+
+def label2hash(meta_json, last):
+    for clss in meta_json['classes']:
+        if clss['title'] == last['classTitle'].values[0]:
+            meta_nodes = clss['geometry_config']['nodes']
+            label2hash = {}
+            for name in meta_nodes:
+                label2hash[meta_nodes[name]['label']] = name
+    return label2hash
+
+
+def fit(imageSize, keypoints_2d, keypoints_3d, focus=1):
+
+    objectPoints = keypoints_3d.loc[keypoints_2d.index].values
+    imagePoints = keypoints_2d[['x', 'y']].values.astype('float')
+
+    n = len(imagePoints)
+    fx = fy = focus*np.hypot(*imageSize)
+    cx = imageSize[1]/2
+    cy = imageSize[0]/2
+    distCoeffs = np.zeros(4, np.float32)
+    if n < 6:
+        raise ValueError('Number of keypoints must be > 5')
+
+    cameraMatrix = np.float32([[fx,0, cx],
+                               [0, fy,cy],
+                               [0, 0, 1]])
+
+    _, rvecs, tvecs = cv.solvePnP(objectPoints, imagePoints, cameraMatrix, distCoeffs, flags=cv.SOLVEPNP_ITERATIVE )
+    return rvecs, tvecs, cameraMatrix, distCoeffs
+
+
+def draw_cloud(img, points_3d, params):
+    imgpts, _ = cv.projectPoints(points_3d, *params)
+
+    for p in imgpts[:, 0]:
+        img = cv.circle(img, p.astype(int), 0, (255,20,147), -1)
+
+
+def draw_keypoints(img, keypoints_3d, params):
+    imgpts, _ = cv.projectPoints(keypoints_3d, *params)
+
+    for p in imgpts[:, 0]:
+        img = cv.circle(img, p.astype(int), 5, (0,0,0), -1)
+    
+    
+def resize(img, width=1000):
+    y, x = img.shape[:2]
+    return cv.resize(img, (width, int(y/x*width)))
+
+
+class Start_annotation():
+    def __init__( self, project_id, dataset_id, token,
+                     local_dataset_path, keypoints_3d_path, point_cloud_path): 
+        
+        adress = 'https://app.supervise.ly/'
+#         self.local_dataset_path = local_dataset_path
+#         self.points_3d = np.asarray(o3d.io.read_point_cloud(point_cloud_path).points)
+        self.keypoints_3d = pd.read_csv(keypoints_3d_path, index_col=0).astype(float)
+
+        self.api = sly.Api(adress, token)
+        self.meta_json = self.api.project.get_meta(project_id)
+        self.meta = sly.ProjectMeta.from_json(self.meta_json)
+        self.images = pd.DataFrame(self.api.image.get_list(dataset_id)).sort_values('name', ignore_index=True)
+    
+
+
+    def load_ann(self, img_id):
+        ann_info = self.api.annotation.download(img_id)
+        return ann_info
+    
+    def annotations(self, ann_info):
+
+        last = last_keypoints_on_img(ann_info)
+        if len(last) == 0:
+            return 
+        nodes = ann_info[2]['objects'][last.index[0]]['nodes']
+        keypoints_2d = pd.DataFrame(columns=['x', 'y'])
+
+        for i in range(1, len(nodes)+1):
+            keypoints_2d.loc[i] = nodes[label2hash(self.meta_json, last)[str(i)]]['loc']
+
+        return keypoints_2d
+    
+    def new_annotations(self, ann_info, new_keypoints):
+
+        last = last_keypoints_on_img(ann_info)
+        nodes = ann_info[2]['objects'][last.index[0]]['nodes']
+
+        for i in new_keypoints.index:
+            nodes[label2hash(self.meta_json, last)[str(i)]]['loc'] = new_keypoints.loc[i].tolist()
+
+        return ann_info
+    
+    def start(self):
+        app = wx.App()
+        wnd = GUI(self.images, self.transform_by_visible)
+        wnd.Show(True)
+        app.MainLoop()
+
+    def transform_by_visible(self, idxs, img_id, name, focus=1, send=True, all_points=False, change_all=False, plot=False):
+        if send==plot==False:
+            return 'Error_empty_request'
+        ann_info = self.load_ann(img_id)
+        keypoints_2d = self.annotations(ann_info)
+        if keypoints_2d is None:
+            return 'Error_annotations'
+        if not all_points:
+            keypoints_2d = keypoints_2d.loc[idxs]
+        imgSize = list(ann_info.annotation['size'].values())
+        params = fit(imgSize, keypoints_2d, self.keypoints_3d, focus)
+        new_keypoints = pd.DataFrame(cv.projectPoints(self.keypoints_3d.values, *params)[0][:, 0],
+                                     columns=['x', 'y'], index=range(1, len(self.keypoints_3d)+1))
+        if not change_all:
+            new_keypoints = new_keypoints.drop(idxs)
+        if send:
+            new_ann = self.new_annotations(ann_info, new_keypoints)
+            new_ann = sly.Annotation.from_json(new_ann.annotation, self.meta)
+            self.api.annotation.upload_ann(img_id, new_ann)
+        if plot:
+            img = cv.imread(os.path.join(self.local_dataset_path, name))
+    #         for p in keypoints_2d.values:
+    #             img = cv.circle(img, p.astype(int), 8, (255,255,255), -1)
+            draw_cloud(img, self.points_3d, params)
+            cv.imshow(name, resize(img, 1200))
+            a = cv.waitKey(0)
+            if (a==ord('q')) | (a==233):
+                cv.destroyAllWindows()
+        return len(keypoints_2d)
+
+class GUI( wx.Frame ):
+    def __init__( self, images, func):
+        wx.Frame.__init__ ( self, None, id = wx.ID_ANY, title = 'SuperviselyKeypointsGui', pos = wx.DefaultPosition, size = wx.Size( 250,450 ), style = wx.CAPTION|wx.CLOSE_BOX|wx.SYSTEM_MENU|wx.RESIZE_BORDER|wx.TAB_TRAVERSAL )
+        self.images = images
+        self.func = func
+        self.SetSizeHints( wx.DefaultSize, wx.DefaultSize )
+
+        gbSizer1 = wx.GridBagSizer( 0, 0 )
+        gbSizer1.SetFlexibleDirection( wx.VERTICAL )
+        gbSizer1.SetNonFlexibleGrowMode( wx.FLEX_GROWMODE_SPECIFIED )
+
+        gbSizer1.SetMinSize( wx.Size( 200,400 ) )
+
+        bSizer3 = wx.BoxSizer( wx.VERTICAL )
+        self.m_checkBoxes = []
+        for i in range(25):
+            CheckBox = wx.CheckBox( self, wx.ID_ANY, u"{}".format(str(i+1)), wx.DefaultPosition, wx.DefaultSize, 0 )
+            self.m_checkBoxes.append(CheckBox)
+            bSizer3.Add( self.m_checkBoxes[i], 0, wx.ALL, 0 )
+
+        gbSizer1.Add( bSizer3, wx.GBPosition( 0, 0 ), wx.GBSpan( 1, 1 ), wx.EXPAND|wx.LEFT|wx.TOP, 5 )
+        wSizer5 = wx.WrapSizer( wx.HORIZONTAL, 0 )
+
+        m_radioBox5Choices = [ u"True", u"False" ]
+        self.radioBox_send = wx.RadioBox( self, wx.ID_ANY, u"SendKeypoints", wx.DefaultPosition, wx.DefaultSize, m_radioBox5Choices, 1, wx.RA_SPECIFY_COLS )
+        self.radioBox_send.SetSelection( 1 )
+        wSizer5.Add( self.radioBox_send, 1, wx.ALL, 5 )
+
+        m_radioBox7Choices = [ u"True", u"False" ]
+        self.radioBox_plot = wx.RadioBox( self, wx.ID_ANY, u"Plot", wx.DefaultPosition, wx.DefaultSize, m_radioBox7Choices, 1, wx.RA_SPECIFY_COLS )
+        self.radioBox_plot.SetSelection( 1 )
+        wSizer5.Add( self.radioBox_plot, 1, wx.ALL, 5 )
+        
+        self.m_checkBox26 = wx.CheckBox(self, wx.ID_ANY, u"All points", wx.DefaultPosition, wx.DefaultSize, 0)
+        wSizer5.Add( self.m_checkBox26, 0, wx.ALL, 5 )
+        
+        self.m_staticText7 = wx.StaticText( self, wx.ID_ANY, u"0", wx.DefaultPosition, wx.DefaultSize, 0 )
+        self.m_staticText7.Wrap( -1 )
+        wSizer5.Add( self.m_staticText7, 0, wx.ALL, 5 )
+
+        self.m_button3 = wx.Button( self, wx.ID_ANY, u"Go", wx.DefaultPosition, wx.DefaultSize, 0 )
+        wSizer5.Add( self.m_button3, 0, wx.ALL, 5 )
+
+        gbSizer1.Add( wSizer5, wx.GBPosition( 0, 1 ), wx.GBSpan( 1, 1 ), wx.ALIGN_CENTER_HORIZONTAL|wx.EXPAND|wx.TOP, 0 )
+
+        m_choice3Choices = self.images.name.tolist()
+        self.m_choice3 = wx.Choice( self, wx.ID_ANY, wx.DefaultPosition, wx.DefaultSize, m_choice3Choices, 0 )
+        self.m_choice3.SetSelection( 0 )
+        gbSizer1.Add( self.m_choice3, wx.GBPosition( 1, 0 ), wx.GBSpan( 2, 8 ), wx.ALL, 5 )
+        
+        self.SetSizer( gbSizer1 )
+        self.Layout()
+
+        self.Centre( wx.BOTH )
+        self.m_button3.Bind( wx.EVT_BUTTON, self.Go )
+
+    def __del__( self ):
+        pass
+    
+    def Go( self, event ):
+        idxs = []
+        for i, checkBox in enumerate(self.m_checkBoxes, start=1):
+            if checkBox.IsChecked():
+                idxs.append(i)
+        all_points = self.m_checkBox26.IsChecked()
+        if (len(idxs)<6) & (not all_points):
+            wx.MessageBox('Точек должно быть больше 5', 'Ошибка', wx.OK)            
+            event.Skip()
+            return            
+        if all_points:
+            n = 25
+        else:
+            n = len(idxs)
+        self.m_staticText7.SetLabel(str(n))
+        img_name = self.m_choice3.GetString(self.m_choice3.GetSelection())
+        send = self.radioBox_send.GetString(self.radioBox_send.GetSelection())
+        plot = self.radioBox_plot.GetString(self.radioBox_plot.GetSelection())
+        img_id = int(self.images[self.images.name==img_name].id.values[0])
+        out = self.func(idxs, img_id, img_name, 1, send=(send=='True'), plot=(plot=='True'), all_points=all_points)  
+        if out == 'Error_empty_request':
+            wx.MessageBox('Выберите plot или send', 'Ошибка', wx.OK) 
+        elif out == 'Error_annotations':
+            wx.MessageBox('На этом изображении нет разметки', 'Ошибка', wx.OK)         
+        event.Skip()
+        return
+    
+    
+if __name__ == '__main__':
+    
+    point_cloud_path = r''
+    keypoints_3d_path = r'karussel_24kps.csv'
+    local_dataset_path = r''
+    token = ''
+    dataset_id = 627375 #(images)
+    project_id = 184347 #(Nurburg-karussel)
+
+    sp = Start_annotation(project_id, dataset_id, token,
+                         local_dataset_path, keypoints_3d_path, point_cloud_path)
+    
+    sp.start()
--- a/Karussell/Supervisely/SuperviselyKeypointsGUI/karussel_24kps.csv
+++ b/Karussell/Supervisely/SuperviselyKeypointsGUI/karussel_24kps.csv
@@ -0,0 +1,25 @@
+,x,y,z
+1,129,-137,0
+2,-41,33,0
+3,-65,46,0
+4,22,-134,0
+5,70,-253,0
+6,69,-329,0
+7,-37,-224,0
+8,-88,-199,0
+9,-143,-171,0
+10,-173,-119,0
+11,-219,-30,0
+12,-307,-26,0
+13,-207,123,0
+14,-138,258,0
+15,-43,240,0
+16,10,242,0
+17,62,229,0
+18,146,174,0
+19,213,138,0
+20,230,59,0
+21,302,-11,0
+22,288,-93,0
+23,172,-65,0
+24,103,-14,0
--- a/Karussell/Supervisely/SuperviselyKeypointsGUI/start.bat
+++ b/Karussell/Supervisely/SuperviselyKeypointsGUI/start.bat
@@ -0,0 +1,4 @@
+@echo on
+call C:\Anaconda\Scripts\activate.bat
+C:\Anaconda\python.exe SuperviselyKeypointsGUI.py
+pause
--- a/Karussell/Supervisely/ToCOCO.ipynb
+++ b/Karussell/Supervisely/ToCOCO.ipynb
@@ -0,0 +1,197 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "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:\\karusel'\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": null,
+   "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": null,
+   "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.astype(float).round().astype(int)\n",
+    "    return keypoints_2d[:24]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "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": null,
+   "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"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "6593f677",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def COCO(image_list, annotations_list):\n",
+    "    coco = {\n",
+    "\n",
+    "        \"info\": {\n",
+    "            \"description\": \"karusel Dataset\", \"version\": \"1.0\"\n",
+    "        },\n",
+    "\n",
+    "        \"categories\": [\n",
+    "            {\n",
+    "                \"supercategory\": \"NurburgRing\",\n",
+    "                \"id\": 1,\n",
+    "                \"name\": \"karusel\",\n",
+    "                \"keypoints\": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11,\n",
+    "                              12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24],\n",
+    "                \"skeleton\": [\n",
+    "                    [1, 2],[2, 3],[3, 4],[4, 5],[5, 6],[6, 7],[7, 8],[8, 9],[9, 10],[10, 11],\n",
+    "                    [11, 12],[12, 13],[13, 14],[14, 15],[15, 16],[16, 17],[17, 18],[18, 19],[19, 20],\n",
+    "                    [20, 21],[21, 22],[22, 23],[23, 24],[24, 1],[24, 3],[1, 5]\n",
+    "                ]\n",
+    "            }\n",
+    "        ]\n",
+    "    }\n",
+    "\n",
+    "    coco['images'] = image_list\n",
+    "    coco['annotations'] = annotations_list\n",
+    "    return coco"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "38880d13",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "coco_json = COCO(*ann_img_list(anns, imgs, meta))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "314565c2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open(os.path.join(path_dataset, 'karusel_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
+}
--- a/Karussell/Supervisely/karusel_create_3d.ipynb
+++ b/Karussell/Supervisely/karusel_create_3d.ipynb
@@ -0,0 +1,102 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9b13894c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import json\n",
+    "import pandas as pd\n",
+    "import numpy as np\n",
+    "import cv2 as cv\n",
+    "import plotly.express as px\n",
+    "import plotly.graph_objects as go\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)]])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3fc452a2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open(r'karusel.png.json', 'r') as file:\n",
+    "    img_json = json.load(file)\n",
+    "    \n",
+    "img = cv.imread(r'karusel.png')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "cfc4d7b9",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df = pd.DataFrame([img_json['objects'][i]['points']['exterior'][0]\\\n",
+    "                   for i in range(len(img_json['objects']))], columns=['x','y'], index=range(1,25))\n",
+    "center = (1165, 874)\n",
+    "df -= center\n",
+    "df[['x', 'y']] = df.values[:, :2]@roration(90) # поворот x и y в плоскости земли\n",
+    "df['z'] = 0\n",
+    "\n",
+    "df = df.loc[[1, 7, 6, 20, 2, 3, 4, 17, 18, 5, 22, 19, 8,\n",
+    "        16, 9, 10, 11, 21, 15, 12, 14, 13, 24, 23]] # упорядочивание индексов\n",
+    "\n",
+    "df.index = list(range(1, 25))\n",
+    "df = df.astype(float)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "07361259",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "img = cv.circle(img, center, 5, (0,0,0), -1)\n",
+    "fig = px.imshow(img)\n",
+    "fig.add_trace(go.Scatter(x=df.x+center[0], y=df.y+center[1], text=df.index,\n",
+    "                         marker=dict(color='red', size=5), mode='markers'))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "4f18693d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df[['x', 'y']] = df.values[:, :2]@roration(44)"
+   ]
+  }
+ ],
+ "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
+}