Assessing High Dynamic Range Imagery Performance for Object Detection in Maritime Environments

<p>The field of autonomous robotics has benefited from the implementation of convolutional neural networks in vision-based situational awareness. These strategies help identify surface obstacles and nearby vessels. This study proposes the introduction of high dynamic range cameras on autonomous surface vessels because these cameras capture images at different levels of exposure revealing more detail than fixed exposure cameras. To see if this introduction will be beneficial for autonomous vessels this research will create a dataset of labeled high dynamic range images and single exposure images, then train object detection networks with these datasets to compare the performance of these networks. Faster-RCNN, SSD, and YOLOv5 were used to compare. Results determined Faster-RCNN and YOLOv5 networks trained on fixed exposure images outperformed their HDR counterparts while SSDs performed better when using HDR images. Better fixed exposure network performance is likely attributed to better feature extraction for fixed exposure images. Despite performance metrics, HDR images prove more beneficial in cases of extreme light exposure since features are not lost.</p>

Name thesis:degree_name

Master of Science in Mechanical Engineering

Level thesis:degree_level

Thesis - Open Access

Discipline thesis:degree_discipline

Mechanical Engineering

Year

2023

Author dc:creator

• Landaeta, Erasmo

• object detection
• situational awareness
• maritime domain
• autonomy
• autonomous surface vessel
• neural networks
• high dynamic range
• Mechanical Engineering
• Ocean Engineering
• Other Mechanical Engineering
• Robotics

Repository record dc:identifier

https://commons.erau.edu/edt/730

OAI identifier oai:identifier

oai:commons.erau.edu:edt-1753