Developing underwater imaging systems and analyzing image datasets with artificial intelligence.
I am an optical and computer vision engineer with a passion to develop novel technology for nature exploration.
During my PhD, I led projects on development of underwater optical instruments that are used to study the biological and physical dynamics of the ocean. My current project as a postdoctoral scholar utilizes a machine learning framework to analyze a large image dataset from an in situ underwater microscope.
A low-cost underwater imaging system for zooplankton studies
An optical instrument for 3D underwater bulkflow estimation
A deep learning framework for analyzing image data sets from an underwater microscope
A laser-pulsed fluorescence and darkfield microscopy for aquatic microogranisms
The In situ Plankton Assemblage eXplorer (IPAX): An inexpensive underwater imaging system for zooplankton study
The IPAX is a low-cost underwater imaging system that can capture images and videos of zooplankton in their natural habitats.
○ Based on the Raspberry Pi environment.
○ Utilizes high-power LEDs with Fresnel lens focusing to illuminate small transparent organisms.
In the second generation, the IPAX is equipped with a stereoscopic vision system with Scheimpflug optics to increase 3D sampling volume.
○ Can be used to localize positions of small aquatic organisms in 3D
○ Can instantaneous measure body length of organisms in free space.
Citation:
Lertvilai, Pichaya. "The In situ Plankton Assemblage eXplorer (IPAX): An inexpensive underwater imaging system for zooplankton study." Methods in Ecology and Evolution 11.9 (2020): 1042-1048.
► The second generation with stereoscopic capability is current under review for publication.
In Situ Underwater Average Flow Velocity Estimation Using a Low-Cost Video Velocimeter
The video velocimeter is a low-cost optical instrument designed to measure 3D velocity of underwater current.
○ It utilizes modified particle image velocimetry technique to measure flow from ambient particles.
○ It is equipped with a unique optical design that split the view of a single camera into a stereoscopic view for 3D reconstruction.
○ The instrument is verified with field experiments to achieve desirable accuracy for underwater flow measurement.
Citation:
Lertvilai, Pichaya, Paul LD Roberts, and Jules S. Jaffe. "In Situ Underwater Average Flow Velocity Estimation Using a Low-Cost Video Velocimeter." Journal of Atmospheric and Oceanic Technology 38.6 (2021): 1143-1156.
A deep learning framework for zooplankton study
A machine learning framework is used to analyze more than a billion images from an underwater microscope system to determine the abundance of zooplankton in Southern California and the environmental factors that drive it.
○ Utilize a RESNET18 architecture with Pytorch framework.
○ Utilize parallel computing to accelerate analyses.
►SPICI is a data management framework to accelerate data handling from server (initially developed by Kevin Le)
►SPC ML is the deep learning framework used in this research to classify zooplankton.
Label-free underwater single-cell and fluorescence imaging of aquatic microorganisms with laser-pulsed darkfield microscopy
The Moore Underwater Microscope is a powerful microscope for studying aquatic microorganisms in situ.
○ Equipped with two high-resolution cameras to capture both fluorescence (450nm excitation) and darkfield (with holographic capability) images at the same time.
○ Generate ~350MB/s of data that are processed by an on-board Jetson TX1 and stored in NVMe drive.
○ Connect with an iPad through an underwater WIFI cable for real-time image feedback underwater for diver deployment.
○ Achieve a detection limit of 1µm cell without using any fluorescence labels.
○ Capable of detect fluorescence of natural pigments in photosynthetic microorganisms.
► The work is in preparation for scientific publication.
Created with
Mobirise.com