Sanghee Kang

About

Sanghee Kang is an Assistant Professor of Second Language Acquisition, English as a Second Language, and Korean Studies at the Department of Languages, Cultures & Applied Linguistics at Carnegie Mellon University. She earned her Ph.D. in Applied Linguistics from Georgia State University, where her dissertation focused on the role of chatbot-based interaction and learner characteristics in the alignment-driven learning of second language grammar and pragmatics. Her research continues to explore the role of generative AI in second language learning, with particular emphasis on AI-generated feedback and collaborative writing. Dr. Kang's interests also include task-based language teaching and digital multimodal composition, targeting both English and Korean as languages of study. She has made significant contributions to Korean Studies, especially in Korean language acquisition among university students, and has been involved in developing research-informed, task-supported Korean language curricula and conducting teacher training workshops on instructional task design. Additionally, she is the co-author of the textbook 'Learning Korean through Tasks: High Beginners to Intermediate,' which integrates recent findings in second language acquisition and task-based language teaching into practical pedagogic tasks. Within the department, she serves as co-coordinator of the Korean Studies program and actively mentors students in an innovative academic environment.

Research topics

• Artificial Intelligence
• Computer Science
• Computer vision
• Computer graphics (images)
• Mathematics
• Optics
• Chemistry
• Physics

Selected publications

• LASER: LAtent SpacE Rendering for 2D Visual Localization

  arXiv (Cornell University) · 2022-04-01 · 1 citations

  preprintOpen access

  We present LASER, an image-based Monte Carlo Localization (MCL) framework for 2D floor maps. LASER introduces the concept of latent space rendering, where 2D pose hypotheses on the floor map are directly rendered into a geometrically-structured latent space by aggregating viewing ray features. Through a tightly coupled rendering codebook scheme, the viewing ray features are dynamically determined at rendering-time based on their geometries (i.e. length, incident-angle), endowing our representati…

  PublisherOA PDFDOI

• SALVe: Semantic Alignment Verification for Floorplan Reconstruction from Sparse Panoramas

  arXiv (Cornell University) · 2024-06-27

  preprintOpen accessSenior author

  We propose a new system for automatic 2D floorplan reconstruction that is enabled by SALVe, our novel pairwise learned alignment verifier. The inputs to our system are sparsely located 360$^\circ$ panoramas, whose semantic features (windows, doors, and openings) are inferred and used to hypothesize pairwise room adjacency or overlap. SALVe initializes a pose graph, which is subsequently optimized using GTSAM. Once the room poses are computed, room layouts are inferred using HorizonNet, and the f…

  PublisherOA PDFDOI

• iBARLE: imBalance-Aware Room Layout Estimation

  arXiv (Cornell University) · 2023-08-29

  preprintOpen accessSenior author

  Room layout estimation predicts layouts from a single panorama. It requires datasets with large-scale and diverse room shapes to train the models. However, there are significant imbalances in real-world datasets including the dimensions of layout complexity, camera locations, and variation in scene appearance. These issues considerably influence the model training performance. In this work, we propose the imBalance-Aware Room Layout Estimation (iBARLE) framework to address these issues. iBARLE c…

  PublisherOA PDFDOI

• U2RLE: Uncertainty-Guided 2-Stage Room Layout Estimation

  arXiv (Cornell University) · 2023-04-17

  preprintOpen accessSenior author

  While the existing deep learning-based room layout estimation techniques demonstrate good overall accuracy, they are less effective for distant floor-wall boundary. To tackle this problem, we propose a novel uncertainty-guided approach for layout boundary estimation introducing new two-stage CNN architecture termed U2RLE. The initial stage predicts both floor-wall boundary and its uncertainty and is followed by the refinement of boundaries with high positional uncertainty using a different, dist…

  PublisherOA PDFDOI

• Graph-CoVis: GNN-based Multi-view Panorama Global Pose Estimation

  arXiv (Cornell University) · 2023-04-26

  preprintOpen accessSenior author

  In this paper, we address the problem of wide-baseline camera pose estimation from a group of 360$^\circ$ panoramas under upright-camera assumption. Recent work has demonstrated the merit of deep-learning for end-to-end direct relative pose regression in 360$^\circ$ panorama pairs [11]. To exploit the benefits of multi-view logic in a learning-based framework, we introduce Graph-CoVis, which non-trivially extends CoVisPose [11] from relative two-view to global multi-view spherical camera pose es…

  PublisherOA PDFDOI

Frequent coauthors

• Richard Szeliski

  39 shared

• Heung‐Yeung Shum

  30 shared

• Long Quan

  Hong Kong University of Science and Technology

  27 shared

• Jingyi Yu

  ShanghaiTech University

  21 shared

• C. Lawrence Zitnick

  19 shared

• Yuguang Li

  First Hospital of Jilin University

  17 shared

• Neel Joshi

  Northeastern University

  15 shared

• Will Hutchcroft

  15 shared

Education

• Ph.D., Applied Linguistics

  Georgia State University

• M.A., Applied Linguistics

  Georgia State University

• B.A.

  Seoul National University

Awards & honors

• GAITAR Fellowship, Eberly Center, Carnegie Mellon University…
• Korean Studies Grant, The Academy of Korean Studies, 2025
• Falk Research Grant, Carnegie Mellon University, 2024

Similar researchers at Carnegie Mellon University

• Cheng Lih-44
• Rémi Adam van Compernolleh-26
• Seth Wienerh-14