http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
비디오 영상을 이용한 3차원 재구성 및 객체 인식 모델 개발
이나혁(Nahyuk Lee),이경택(Kyungtaek Lee),박영섭(Youngsup Park),서상현(Sanghyun Seo),이태민(Taemin Lee) 한국디지털콘텐츠학회 2020 한국디지털콘텐츠학회논문지 Vol.21 No.11
As the content industries developed, the scope of content used expanded from two to three dimensions, and not only experts but also ordinary users wanted to create and use this content. But handling three-dimensional information requires a lot of technology and time. Therefore, this study presents a simple three-dimensional reconstruction method using SfM. When users produce video images in a simple way, datasets are augmented and increased based on this, and then carry out three-dimensional reconstruction using the increased data. It also produces models that recognize three-dimensional objects through CNN learning. In other words, we produced two different results, one dataset and one information extraction process.
기계학습 모델 기반의 사진으로부터 감성 분석에 관한 연구
이나혁(Nahyuk Lee),이태민(Taemin Lee) 한국디지털콘텐츠학회 2021 한국디지털콘텐츠학회논문지 Vol.22 No.8
Humans are generally visually dependent animals. Information coming into the eye affects humans the most. Through the development of the Internet, more images are easily accessible through online rather than just experience. Predicting what sensitivity a given image can give users, it can be a little more helpful when users search or categorize images. While more studies have been conducted to predict sensitivity in content as machine learning advances, analysis of which of the machine learning models is effective in predicting emotions has been insufficient. This study is conducted to predict sensitivity based on color information of images, which is visual information. Categorize the 4 representative emotions and crawl the images into one of the four emotions. At this time, through various machine learning, we find the model that best suits each model, and analyze the results of each machine learning model. This finally summarizes whether it is reasonable to analyze images based on color.