미래가치 창출을 위한 수리과학 선도인재 양성팀

일시: 2023년 11월 07일 (화) 16:00 ~ 17:00

장소: 자1204호

강사: 차재훈 (Science and Technology Facilities Council, United Kingdom)

주제: Centroid- and Orientation-aware Feature Learning

초록: Robust techniques for learning the centroids and orientations of objects in two-dimensional images, along with other features, are crucial for image processing applications. In this talk, we present a novel encoder-decoder-based mechanism for learning independent factors of variation, including centroids and orientations, by embedding special layers to achieve translational and rotational equivariance and invariance. The evaluation demonstrates that the proposed model can offer superior performance in learning these centroids, orientations, and other features.