Skip to content

Commit

Permalink
kitti processing with hdbscan notebook
Browse files Browse the repository at this point in the history
  • Loading branch information
@SVivdich02
SVivdich02 committed Apr 13, 2024
1 parent eb703c8 commit 95f9512
Showing 1 changed file with 342 additions and 0 deletions.
342 changes: 342 additions & 0 deletions main_hdbscan_processing.ipynb
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,342 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import copy\n",
"import numpy as np\n",
"import open3d as o3d\n",
"import random\n",
"\n",
"def generate_random_colors(N):\n",
" colors = [[0, 0, 0]]\n",
" for _ in range(N):\n",
" colors.append(\n",
" [random.randint(0, 255), random.randint(0, 255), random.randint(0, 255)]\n",
" )\n",
"\n",
" colors = np.vstack(colors) / 255\n",
" return colors\n",
"\n",
"def color_pcd_by_labels1(pcd, labels):\n",
" colors = generate_random_colors(len(labels) + 1)\n",
" pcd_colored = copy.deepcopy(pcd)\n",
" pcd_colored.colors = o3d.utility.Vector3dVector(\n",
" np.zeros(np.asarray(pcd.points).shape)\n",
" )\n",
"\n",
" for i in range(len(pcd_colored.points)):\n",
" if labels[i] == 0 or labels[i] == -1:\n",
" continue\n",
" pcd_colored.colors[i] = colors[labels[i]]\n",
"\n",
" return pcd_colored"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import pickle\n",
"\n",
"start_num = 20\n",
"end_num = start_num + 4\n",
"\n",
"file_name = \"experiment_bin_0704_4_sem_offset0_T0l03/start{}_end{}.pickle\".format(start_num, end_num)\n",
"\n",
"with open(file_name, 'rb') as file:\n",
" data = pickle.load(file)\n",
"\n",
"print(data[0]) # значения гиперпараметров запуска"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from sklearn.cluster import HDBSCAN\n",
"\n",
"clusterer = HDBSCAN()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"import open3d as o3d\n",
"\n",
"X = np.asarray(data[1]) # точки обработанного облака ДО вокселизации\n",
"pcd_hdbscan = o3d.geometry.PointCloud()\n",
"pcd_hdbscan.points = o3d.utility.Vector3dVector(X)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import copy\n",
"\n",
"# проведение точно такой же вокселизации, как и в моем алгоритме\n",
"\n",
"pcd_hdbscan_copy = copy.deepcopy(pcd_hdbscan)\n",
"\n",
"min_bound = pcd_hdbscan_copy.get_min_bound()\n",
"max_bound = pcd_hdbscan_copy.get_max_bound()\n",
"\n",
"downpcd_trace = pcd_hdbscan_copy.voxel_down_sample_and_trace(\n",
" 0.25, min_bound, max_bound, True\n",
")\n",
"\n",
"downpcd = downpcd_trace[0]\n",
"trace_hdbscan = downpcd_trace[2]\n",
"\n",
"print(\"voxel_down_pcd_size={}\".format(len(downpcd.points)))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# проверяем, что вокселизация привела облако в то состояние,\n",
"# при котором проводилась сегментация моим алгоритмом\n",
"\n",
"trace_arrays = []\n",
"for int_vector in trace_hdbscan:\n",
" trace_arrays.append(np.asarray(int_vector))\n",
"\n",
"for ind, el in enumerate(trace_arrays):\n",
" if (el != data[2][ind]).any():\n",
" print(\"ERROR\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# сегментация HDBSCAN\n",
"\n",
"clusters = clusterer.fit_predict(np.asarray(downpcd.points))\n",
"print(set(clusters))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import copy\n",
"\n",
"from src.utils.pcd_utils import visualize_pcd\n",
"\n",
"colored_pcd = color_pcd_by_labels1(copy.deepcopy(downpcd), clusters)\n",
"visualize_pcd(colored_pcd)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# приводим результаты кластеризации к листу, элементы которого - массивы точек одного кластера\n",
"\n",
"clusters_list = []\n",
"\n",
"print(\"before set(clusters)={}\".format(set(clusters)))\n",
"# метки 0 и -1 означают шум, они для нас равны\n",
"for ind, label in enumerate(clusters):\n",
" if label == -1:\n",
" clusters[ind] = 0\n",
"print(\"after set(clusters)={}\".format(set(clusters)))\n",
"\n",
"# инициализация листа\n",
"for label in set(clusters):\n",
" clusters_list.append([])\n",
"\n",
"for ind, label in enumerate(clusters):\n",
" clusters_list[label].append(ind)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import copy\n",
"\n",
"from src.utils.pcd_utils import color_pcd_by_clusters_and_voxels\n",
"from src.utils.pcd_utils import visualize_pcd\n",
"\n",
"colored_pcd = color_pcd_by_clusters_and_voxels(\n",
" copy.deepcopy(pcd_hdbscan_copy), copy.deepcopy(trace_hdbscan), clusters_list\n",
")\n",
"visualize_pcd(colored_pcd)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# выборка из кластеризации только инстансов\n",
"\n",
"def find_num_in_inst_label_array(src_points, inst_label_array_for_clustering):\n",
" for point in src_points:\n",
" if inst_label_array_for_clustering[point] > 0:\n",
" return inst_label_array_for_clustering[point]\n",
" return -1\n",
"\n",
"\n",
"def build_pred_inst_array(\n",
" inst_label_array_for_clustering, clusters, trace, instance_threshold\n",
"):\n",
" pred_inst_array = np.zeros(len(inst_label_array_for_clustering), dtype=int)\n",
" free_id = 1\n",
" for cluster in clusters:\n",
" voxel_not_in_gt_cluster_count = 0\n",
" for voxel in cluster:\n",
" src_points = trace[voxel]\n",
" id = find_num_in_inst_label_array(\n",
" src_points, inst_label_array_for_clustering\n",
" )\n",
" if id == -1:\n",
" voxel_not_in_gt_cluster_count += 1\n",
"\n",
" cluster_in_gt_instance = (\n",
" (len(cluster) - voxel_not_in_gt_cluster_count) / len(cluster)\n",
" ) * 100\n",
" if cluster_in_gt_instance >= instance_threshold:\n",
" for voxel in cluster:\n",
" src_points = trace[voxel]\n",
" for src_point in src_points:\n",
" pred_inst_array[src_point] = free_id\n",
" free_id += 1\n",
" return pred_inst_array"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"instance_threshold = 30\n",
"inst_label_array_for_clustering = data[4] # из pickle\n",
"\n",
"pred_inst_array = build_pred_inst_array(\n",
" copy.deepcopy(inst_label_array_for_clustering),\n",
" clusters_list,\n",
" copy.deepcopy(trace_hdbscan),\n",
" instance_threshold,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from evops.metrics import precision\n",
"from evops.metrics import recall\n",
"from evops.metrics import fScore\n",
"\n",
"pred_labels = pred_inst_array\n",
"gt_labels = inst_label_array_for_clustering\n",
"tp_condition = \"iou\"\n",
"print(\"precision={}\".format(precision(pred_labels, gt_labels, tp_condition)))\n",
"print(\"recall={}\".format(recall(pred_labels, gt_labels, tp_condition)))\n",
"print(\"fScore={}\".format(fScore(pred_labels, gt_labels, tp_condition)))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import copy\n",
"\n",
"from src.utils.pcd_utils import color_pcd_by_labels\n",
"from src.utils.pcd_utils import visualize_pcd\n",
"\n",
"colored_clusters_for_clustering = color_pcd_by_labels(\n",
" copy.deepcopy(pcd_hdbscan_copy), pred_labels\n",
")\n",
"visualize_pcd(colored_clusters_for_clustering)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import copy\n",
"\n",
"from src.utils.pcd_utils import color_pcd_by_labels\n",
"from src.utils.pcd_utils import visualize_pcd\n",
"\n",
"colored_clusters_for_clustering = color_pcd_by_labels(\n",
" copy.deepcopy(pcd_hdbscan_copy), gt_labels\n",
")\n",
"visualize_pcd(colored_clusters_for_clustering)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"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": "3.9.6"
},
"orig_nbformat": 4
},
"nbformat": 4,
"nbformat_minor": 2
}

0 comments on commit 95f9512

Please sign in to comment.