Autonomous driving involves utilizing various sensors to perceive the surrounding environment and control the vehicle. Camera sensors, depending on their placement, type, and driving conditions, can exhibit disparities between images known as Domain Gap. Previous research has extensively applied Unsupervised Domain Adaptation techniques to overcome performance degradation due to photometry and texture gaps in images. However, most existing studies neglect the optical characteristics of cameras, especially geometric distortion in images.
Therefore, in this competition, we propose the development of an AI algorithm capable of achieving high-performance image segmentation (Semantic Segmentation) for distorted images (Target Domain). This will be accomplished by leveraging undistorted images (Source Domain) and their corresponding labels.

We selected 'SePiCo: Semantic-Guided Pixel Contrast for Domain Adaptive Semantic Segmentation' as a method of Domain Adaptation

Team: Hyunjae Kim

스마트 공장의 제어 시스템 구축을 위한 제품 품질 분류 AI 모델 개발

Team: Yunah Choi

Implemented a Game Review NLP Dashboard applying NLP methods such as BiLSTM and Fine-tuned KoBERT. The project received the Grand Prize at the 2023 Spring INSIGHT Projects Competition.

Developed the main page configuration using Python, Pandas, NumPy, PyTorch, TensorFlow, Docker, FastAPI, KoNLPy, and Plotly.

Team: Hyunjae Kim, Chanjoong Lee, Beomyun Kwon, Chanho Park, Minhye Lee

Developed a Risk Rating model for small business owners in Gangseo-gu using the credit rating model. Received the Mayor of Gangseo-gu Office Award.

Conducted data analysis and processing, and designed a slide using Python, Pandas, NumPy, and Tableau.

Team: Heejin Park, Mijung Bang

Developed a Carbon Footprint calculator and Eco-friendly travel route recommendation system in Jeju Island. Awarded the Innovation League Grand Prize (Runner-Up) at BIG CONTEST '22.

Utilized Streamlit for deployment, applied NLP methods, and designed user-specific ML algorithms.

Team: Jeonghyun Ha, Kyunghoon Na, Kyungjoo Ko, Hannah Jung

Designed a custom shell that translates user's commands to the Kernel using C language. Built Linux shell with basic commands like 'cd', 'mkdir', 'rmdir', etc.

Created child processes with the 'fork()' function and implemented a parent process to create new child processes for each instruction.

Constructed a concurrent stock server that can be connected by multiple clients for buying or selling stocks concurrently.

Implemented two approaches: Event-based and Thread-based. Displayed the status of the current stock written from the stock.txt file.

Existing water photovoltaic power generation uses the sun's semi-permanent energy and water, an idle resource. And this can only be applied on the water without waves. Therefore, we built a water-tracking solar power generation system.

Laser cutting machines can easily process materials, but they are heavy and have the disadvantage that the size of the material is limited depending on the size of the laser cutting machine. Our team wants to compensate for the shortcomings of the laser cutting machine by using the Mechanum wheel. The driving equation and efficiency according to the direction of movement of the mechanical wheel are explored. It also develops a device that cuts the material by moving it according to the drawing.