訊息公告

【PAIR中心AI meetup活動資訊】6月10日(一) 上午11:00~下午1:00,主題:Domain Adaptation for Semantic Segmentation / Joint Learning of Stereo Matching and Optical Flow

【PAIR中心AI meetup活動公告】

 

人工智慧普適研究中心轄下計畫- 交通大學莊仁輝教授團隊將於2019年6月10日星期一舉辦AI Meetup,歡迎有興趣的師生、產業主管、研究人員及工程師等踴躍報名參加,機會難得,請不要錯過!! 

 

(歡迎轉知分享此活動訊息)

 

時  間:2019年6月10日星期一 上午11:00~下午1:00
地  點:國立交通大學工程三館345會議室 (新竹市大學路1001號)
演 講 者:邱維辰助理教授 -國立交通大學資訊工程學系
演講主題:Domain Adaptation for Semantic Segmentation / Joint Learning of Stereo Matching and Optical Flow
演講摘要:

In this meetup we sequentially discuss two topics: (1) unsupervised learning of stereo matching and optical flow, and (2) unsupervised domain adaptation for the task of semantic segmentation.
First, stereo matching and optical flow are two essential tasks for scene understanding, spatially in 3D and temporally in motion. We investigate the strong connection between these two tasks and propose a single and principled network to jointly learn spatiotemporal correspondence for stereo matching and flow estimation, with a newly designed geometric connection as the unsupervised signal for temporally adjacent stereo pairs. Second, for the task of semantic segmentation with unsupervised domain adaptation, we attempt to transfer the knowledge learned upon synthetic datasets with ground-truth labels to real-world images without any annotation. With the hypothesis that the structural content of images is the most decisive factor to semantic segmentation, we propose a Domain Invariant Structure Extraction (DISE) framework to disentangle images into domain-invariant structure and domain-specific texture representations, and obtain superior performance.



主辦單位:科技部人工智慧普適研究中心
協辦單位:國立交通大學電腦視覺研發中心
活動洽詢:林小姐 03-5712121#52991 lsp608@g2.nctu.edu.tw
費  用:免費報名參加
報名對象:對AI領域有興趣的教師、學生、產業主管、研究人員及工程師等,皆可報名。

報名網址:https://reurl.cc/YGE6D
注意事項:
(1)參加活動之成員,需同意將活動進行中的照片授權給主辦單位使用。
(2)現場提供餐點,敬請詳填用餐需求,以便統計數量。
(3)主辦單位有修改活動內容之權利。