<?xml version="1.0" encoding="UTF-8"?><?xml-stylesheet type="text/xsl" href="static/style.xsl"?><OAI-PMH xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/ http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd"><responseDate>2026-11-22T16:05:15Z</responseDate><request verb="GetRecord" identifier="oai:uwspace.uwaterloo.ca:10012/14720" metadataPrefix="dim">https://dspacemainprd01.lib.uwaterloo.ca/server/oai/request</request><GetRecord><record><header><identifier>oai:uwspace.uwaterloo.ca:10012/14720</identifier><datestamp>2025-03-26T23:37:37Z</datestamp><setSpec>com_10012_9900</setSpec><setSpec>com_10012_1</setSpec><setSpec>col_10012_9912</setSpec><setSpec>col_10012_6</setSpec></header><metadata><dim:dim xmlns:dim="http://www.dspace.org/xmlns/dspace/dim" xmlns:doc="http://www.lyncode.com/xoai" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.dspace.org/xmlns/dspace/dim http://www.dspace.org/schema/dim.xsd">
   <dim:field mdschema="dc" element="contributor" qualifier="advisor">Waslander, Steven</dim:field>
   <dim:field mdschema="dc" element="contributor" qualifier="author" authority="6f972a0d-e346-4906-a373-2670ccfc75ae" confidence="600">Lee, Jungwook</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="accessioned">2019-05-24T19:59:20Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="available">2019-05-24T19:59:20Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="issued">2019-05-24</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="submitted">2019-05-15</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="uri">http://hdl.handle.net/10012/14720</dim:field>
   <dim:field mdschema="dc" element="description" qualifier="abstract" lang="en">Training 3D object detectors on publicly available data has been limited to small datasets&#xd;
due to the large amount of effort required to generate annotations. The difficulty of labeling&#xd;
in 3D using 2.5D sensors, such as LIDAR, is attributed to the high spatial reasoning skills&#xd;
required to deal with occlusion and partial viewpoints. Additionally, the current methods&#xd;
to label 3D objects are cognitively demanding due to frequent task switching. Reducing&#xd;
both task complexity and the amount of task switching done by annotators is key to&#xd;
reducing the effort and time required to generate 3D bounding box annotations. We&#xd;
therefore seek to reduce the burden on the annotators by leveraging existing 3D object&#xd;
detectors using deep neural networks.&#xd;
&#xd;
This work introduces a novel ground truth generation method that combines human&#xd;
supervision with pre-trained neural networks to generate per-instance 3D point cloud seg-&#xd;
mentation, 3D bounding boxes, and class annotations. The annotators provide object&#xd;
anchor clicks which behave as a seed to generate instance segmentation results in 3D. The&#xd;
points belonging to each instance are then used to regress object centroids, bounding box&#xd;
dimensions, and object orientation. The deep neural network model used to generate the&#xd;
segmentation masks and bounding box parameters is based on the PointNet architecture.&#xd;
&#xd;
We develop our approach with reliance on the KITTI dataset to analyze the quality&#xd;
of the generated ground truth. The neural network model is trained on KITTI training&#xd;
split and the 3D bounding box outputs are generated using annotation clicks collected&#xd;
from the validation split. The validation split of KITTI detection dataset contains 3712&#xd;
frames of pointcloud and image scenes and it took 16.35 hours to label with the following&#xd;
method. Based on these results, our approach is 19 times faster than the latest published&#xd;
3D object annotation scheme. Additionally, it is found that the annotators spent less&#xd;
time per object as the number of objects in the scenes increase, making it a very efficient&#xd;
for multi-object labeling. Furthermore, the quality of the generated 3D bounding boxes,&#xd;
using the labeling method, is compared against the KITTI ground truth. It is shown that&#xd;
the model performs on par with the current state-of-the-art 3D detectors and the labeling&#xd;
procedure does not negatively impact the output quality of the bounding boxes. Lastly, the&#xd;
proposed scheme is applied to previously unseen data from the Autonomoose self-driving&#xd;
vehicle to demonstrate generalization capabilities of the network.</dim:field>
   <dim:field mdschema="dc" element="language" qualifier="iso" lang="en">en</dim:field>
   <dim:field mdschema="dc" element="publisher" lang="en">University of Waterloo</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Machine Learning</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Computer Vision</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Autonomous Driving</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Deep Learning</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Object Detection</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en">Data Mining</dim:field>
   <dim:field mdschema="dc" element="title" lang="en">3D Ground Truth Generation Using Pre-Trained Deep Neural Networks</dim:field>
   <dim:field mdschema="dc" element="type" lang="en">Master Thesis</dim:field>
   <dim:field mdschema="dc" element="pending">false</dim:field>
   <dim:field mdschema="uws-etd" element="degree" qualifier="department" lang="en">Mechanical and Mechatronics Engineering</dim:field>
   <dim:field mdschema="uws-etd" element="degree" qualifier="discipline" lang="en">Mechanical Engineering</dim:field>
   <dim:field mdschema="uws-etd" element="degree" qualifier="grantor" lang="en">University of Waterloo</dim:field>
   <dim:field mdschema="uws-etd" element="degree" lang="en">Master of Applied Science</dim:field>
   <dim:field mdschema="uws" element="contributor" qualifier="advisor">Waslander, Steven</dim:field>
   <dim:field mdschema="uws" element="contributor" qualifier="affiliation1" lang="en">Faculty of Engineering</dim:field>
   <dim:field mdschema="uws" element="published" qualifier="city" lang="en">Waterloo</dim:field>
   <dim:field mdschema="uws" element="published" qualifier="country" lang="en">Canada</dim:field>
   <dim:field mdschema="uws" element="published" qualifier="province" lang="en">Ontario</dim:field>
   <dim:field mdschema="uws" element="typeOfResource" lang="en">Text</dim:field>
   <dim:field mdschema="uws" element="peerReviewStatus" lang="en">Unreviewed</dim:field>
   <dim:field mdschema="uws" element="scholarLevel" lang="en">Graduate</dim:field>
   <dim:field mdschema="others" element="access-status">open.access</dim:field>
</dim:dim>
</metadata></record></GetRecord></OAI-PMH>