Research

We introduce an accurate eye-tracking method that exploits deflectometric information and uses deep learning to reconstruct the gaze direction. We demonstrate real world experiments with evaluated gaze errors below 1°.

Proposed a novel method for the digital preservation of analog film holograms. We also evaluate the performance of NeRF hologram preservation method with both qualitative and quantitative experiments.

[ Read More ]

Proposed a way of preserving holography images through rendering the holography images with NeRF and NeRF--, a neural radiance field representation method. Also measures the feasibility of interpolating photographs of holograms by using the neural radiance field to synthesize novel views. Through quantitative analysis, it demonstrates that the NeRF model is able to interpolate photographs of holograms.

[ Read More ]

Proposed a machine learning based approach to classify compression fracture and scoliosis spinal images, and proposed novel DICOM image enhancing windowing algorithm.

• [ Nov 2021 ] The proposed classification model is in an installation process to the multiple affiliated hospitals in nationwide (South Korea), including Seoul National University Hospital (SNUH).

• [ Dec 2022 ] The proposed DICOM file windowing algorithm and the classification model are adapted to the deep-learning-based algorithm that measures the compression ratio (CR) for compression fracture (CF) diagnosis in lateral X-rays. This work was presented at Radiological Society of North America (RSNA 2022).

[ Read More ]

Proposed a NetLogo extension for cartesian genetic programming, and designed multi-agent modeling environment which agents may collaborate with one another.

[ Read More ]

Perform autonomous construction in unstructured terrain, and designed a robotic system for a four wheeled-robot to autonomously modify a terrain to build motion support structures.

[ Read More ]

Compared the performance of a four-wheeled robot with AprilTag and Intel RealSense Tracking Camera T265 in 4 different types of terrain.

[ Read More ]

Perform a parameter tuning to make the robot search for speed over accuracy. With a thorough analysis, we choose a value that does not “deviate” greatly from the best setting, and has considerably lower run times.

[ Read More ]

Proposed an algorithmic approach for autonomous construction from a collection of irregularly shaped objects. These plans are executed open-loop with a robotic arm (UR5) equipped with a wrist RGB-D camera.

[ Read More ]