Sökning: "3D Object Detection"
Visar resultat 1 - 5 av 69 uppsatser innehållade orden 3D Object Detection.
1. Instance segmentation using 2.5D data
Master-uppsats, Linköpings universitet/Institutionen för systemteknikSammanfattning : Multi-modality fusion is an area of research that has shown promising results in the domain of 2D and 3D object detection. However, multi-modality fusion methods have largely not been utilized in the domain of instance segmentation. LÄS MER
2. Visual Bird's-Eye View Object Detection for Autonomous Driving
Master-uppsats, Linköpings universitet/DatorseendeSammanfattning : In the field of autonomous driving a common scenario is to apply deep learningmodels on camera feeds to provide information about the surroundings. A recenttrend is for such vision-based methods to be centralized, in that they fuse imagesfrom all cameras in one big model for a single comprehensive output. LÄS MER
3. LiDAR Perception in a Virtual Environment Using Deep Learning : A comparative study of state-of-the-art 3D object detection models on synthetic data
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)Sammanfattning : Perceiving the environment is a crucial aspect of autonomous vehicles. To plan the route, the autonomous vehicle needs to be able to detect objects such as cars and pedestrians. This is possible through 3D object detection. However, labeling this type of data is time-consuming. LÄS MER
4. Robust Multi-Modal Fusion for 3D Object Detection : Using multiple sensors of different types to robustly detect, classify, and position objects in three dimensions.
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)Sammanfattning : The computer vision task of 3D object detection is fundamentally necessary for autonomous driving perception systems. These vehicles typically feature a multitude of sensors, such as cameras, radars, and light detection and ranging sensors. LÄS MER
5. Arbitrary motion Synthetic Aperture Radar
Master-uppsats, Lunds universitet/Matematisk statistikSammanfattning : Syftet med denna avhandling är att utveckla en ny metod för att producera bilder med syntetisk aperturradar (SAR), med utgångspunkt i scenarier med arbiträr rörelse vad gäller radarsensorn. SAR är en väletablerad metod för att skapa 2- eller 3-dimensionella radarbilder, som traditionellt sett antar att radar-sensorns rörelse är linjär och förutsägbar. LÄS MER