Research Summary


Research Interests



1. What are the fundamental sources that make the morphological differences in ants we are looking at? How muscle structures derived from each ants' ecology or behavior can be affected to the ant morphology? Functional morphology
2. How do reproductive behaviors (or ecological facets) lead to heterogeneous genital structures in male ants and how we can delimitate the boundaries of ant species especially using reproductive castes? - Taxonomy, Phylogeny, Morphology 
3. How are Aphaenogaster ants distributed in across the world, and where do they come from? - Paleoentomology, Phylo-biogeography
4. What are the potential ant species which can be spread to the world rapidly and what ecological features do they have? - Ecology
5. Could taxonomy be enhanced by using deep learning and Computer vision? - Deep learning, Computer vision

1. 3D based ant morphology and anatomy

Taxonomy, Systematics, Morphology

 

 


3D BASED MORPHOLOGY

Figuring out the clear morphological characteristics are still considered highly important in the taxonomy even though molecular studies based on various markers are getting more important to understand the taxonomic status of each species. 3D modeling can be a very useful method when it can be properly provided with CT-scanned stl file or many 2D scientific photographs or SEM images. It is also can be freely sent as a 3D file format and printed out. I expect that I might be able to get some 3D morphometrics data based on these 3D models. All the 3D files can be further animated by Autodesk MAYA and some of rendering programs to visualize the specific behaviors such as walking patterns or mating behaviors.

L: 3D model of Neoponera sp.

1. 3D model of Linguamyrmex sp.

2. 3D model of Neoponera sp. 

3. 3D model of Aphaenogaster sp,

4. 3D model of Aphaenogaster sp, (2)


‌ A photogrammetry is very helpful method in examining the particular characters which are might very important in taxonomy but hard to visible in 2D photos. It can bring the micro habitats of each species to virtual reality as well. It will be way more informative than traditional biological descriptions. ‌ 

 

2. Using deep learning and Computer vision in entomology

 Image segmentation, object detecting and motion tracking

Automatic ant ID:
Image segmentation, multiple object detecting and motion tracking

Deep learning and Computer vision are becoming very strong tools to solve the problems in many scientific fields. I can use some of basic skills for constructing deep learning which are represented by Pytorch and Tensorflow. Image classification by R-CNN algorithm, real-time multiple objects detection,  and Image segmentation will be very attractive ways to do many entomological studies that we could not been approached. 

L: Automatic ant ID program I made using Tensorflow. Please wait for a second to see how it works

1. Vehicle detecting program I have conducted using Pytorch and YOLO v5 (1)

2. Tensorboard to see how the program I have made works well 

3. Vehicle detecting program I have conducted using Pytorch and YOLO v5 (2)

4. Image crawling using Selenium Python to make my own dataset .

3. Ant Phylogeny & Evolutionary biology

Paleomyrmecology, Impression and amber fossils

 

 

1. Impression fossil excavated from Pohang

2. Impression fossil excavated from Jinju

3. Line drawing of impression fossil

4. Co-evolution: ants and other myrmecophile animals

The relationships between ants and other creatures