CS Events Monthly View
Qualifying ExamLeveraging Powerful Attention Mechanisms for Biological Image Segmentation |
|
||
Thursday, April 27, 2023, 04:00pm - 06:00pm |
|||
Abstract: The application of advanced attention mechanisms, originally designed for natural language processing and computer vision, has consistently delivered outstanding performance in diverse tasks across different data modalities. Its adaptability, long-range dependency capturing ability, and good parallelism capability make it well-suited for enhancing the accuracy and completeness of automated image analysis, particularly in the domain of biological image segmentation. In this presentation, we will delve into two innovative approaches, both of which leverage attention mechanisms to effectively tackle the challenges associated with merging multi-view imaging information and improving segmentation mask accuracy and completeness. These works showcase novel strategies for addressing the complexities of automated image analysis, with the ultimate objective of elevating diagnostic accuracy and efficiency. Experiment results across multiple datasets corroborate the effectiveness of our models in improving image segmentation quality.
Speaker: Qilong Zhangli
Location : CoRE 305
Committee:
Professor Dimitris Metaxas (Chair)
Professor Konstantinos Michmizos
Professor Yongfeng Zhang
Professor Aaron Bernstein
Event Type: Qualifying Exam
Abstract: See above
Organization:
Rutgers University
School of Arts & Sciences
Department of Computer Science
Contact Professor Dimitris Metaxes