- 最后登录
- 2017-7-6
- 在线时间
- 81 小时
- 寄托币
- 207
- 声望
- 0
- 注册时间
- 2008-7-5
- 阅读权限
- 20
- 帖子
- 10
- 精华
- 0
- 积分
- 198
- UID
- 2511674
- 声望
- 0
- 寄托币
- 207
- 注册时间
- 2008-7-5
- 精华
- 0
- 帖子
- 10
|
本帖最后由 tshxuan 于 2017-4-27 14:38 编辑
请大家帮忙扩散一个新加坡南阳理工大学(NTU)招生的消息。我是老师以前的学生,老师人很nice,六月份会去NTU入职,以前在新加坡研究所(Institute for Infocomm Research)任职senior research scientist,主要做计算机视觉,机器学习和深度学习方面的研究。奖学金刚开始是每个月2k新币(~1w RMB),第二年通过一个考核后会提升到2.5k,奖学金持续四年,一般都可以在这段时间毕业。毕业后不会被强制留下来在新加坡工作,去留随意。希望可以找到国内985,211院校的优秀学生一起做研究。有意向的同学可以发邮件联系:Shijian.Lu@ntu.edu.sg
以下是详细信息:
NTU PhD Research Scholarship
Eligibility criteria
• You must have a First Class Honours or Second Class (Upper Division) Honours or its equivalent
• You should not be on paid employment or accept paid employment or concurrently hold any other
scholarship, fellowship, bursary or top-up allowance during the prescribed period of the award
Coverage
• The award is tenable for one year in the first instance and is renewable subject to good progress.
• The monthly stipend will be S$2,000 (around ¥ 10, 000) per month. After passing the Ph.D.
Qualifying Examination, the stipend will be increased to S$2,500 (around ¥ 12, 500), subject to good performance in research and the attainment of required standards for courses taken.
• The award also covers the annual tuition fee and the annual computer fee.
• The maximum period of the award is up to four years for Ph.D. candidates.
Research work and duties
• You must undertake to carry out research work at NTU.
• Upon your confirmation of PhD candidature, you will be required to assist your School in teaching duties for three hours a week. There will be no remuneration.
• In addition, the School, at its discretion, may appoint you to assist in academic or administration work not exceeding 10 hours of work per week. There may or may not be any remuneration.
Tentative Topic: Generic Visual Search based on Deep Learning and Attention Model
Object detection has played quite important role in building intelligent robot in interacting with people
and environment. Recently, important progress for improving the accuracy of object detectors has been
made possible with Convolutional Neural Networks (CNNs), which leverage big visual data and deep
learning for image categorization. While these techniques focused on still images, determining the exact
location of a target object in a scene requires active engagement to understand the context, change the
fixation point, identify distinctive parts that support recognition, and determine the correct proportions
of the box, as conveyed in sequential video stream.
The goal of this project is to develop accurate and efficient visual target search models for video streams
like movies or even challenger RGBD streams shot by Kinect. Advanced visual attention and deep
network learning techniques will be designed for the visual target search. The developed technique will
also be evaluated for other large scale video understanding tasks such as video object detection,
classification, retrieval and description. The project will target publications in reputable conferences and
journals such as ICCV, CVPR, and TPAMI. In addition, the project will develop demonstrable prototype
systems by working with other members within a research team.
Contact
This PhD research will be conducted under the supervision of Asst Prof Lu Shijian within the School of
Computer Science and Engineering, Nanyang Technological University. Interested candidates please
contact him at Shijian.Lu@ntu.edu.sg. |
|