HKBUtube

Distinguished lecture (Hong Kong Baptist University. Dept. of Computer Science)

2210 Views 觀看次數

Open to All 公開

00:00

Recommended Videos 影片推薦

Audio 語言

Who Is There? Face Recognition Applications

DEPARTMENT:

Department of Computer Science

CONYTRIBUTOR / SPEAKER:

JAIN, Anil

RELEASE DATE:

2015

LENGTH:

70 Minutes

VIDEO TYPE:

Scholarly Talks

There are two main drivers of face recognition technology: (i) security, namely, access to restricted areas and personal devices, de-duplication of passports and driver licenses and identifying suspects in surveillance videos, and (ii) social media, where face recognition is useful for automatic tagging of photos. These applications and their requirements have generated tremendous interest in face recognition research and development. While the origins of machine face recognition date back 50 years, the general problem of face recognition is incredibly difficult due to large intra-person face variability: pose, illumination, expression, occlusion, and aging. In other words, different face images of the same person acquired at different times and under different imaging conditions can have quite different appearances that are difficult to match by state of the art systems. Hence, the challenge is to design salient feature extractors and robust matchers for face images. For this reason, convolution neural networks, also known as, deep networks, have played a major role in the new generation of face recognition systems. This talk will address a number of ongoing research projects in my laboratory that include (i) face identification at scale (gallery of 80 million faces), (ii) large-scale face clustering (120 million faces), (iii) longitudinal study of face recognition, and (iv) detection of spoof faces.

Who Is There? Face Recognition Applications

部門:

計算機科學系

講者:

JAIN, Anil

發佈日期:

2015

影片長度:

70 分鐘

影片類型:

學術論壇

There are two main drivers of face recognition technology: (i) security, namely, access to restricted areas and personal devices, de-duplication of passports and driver licenses and identifying suspects in surveillance videos, and (ii) social media, where face recognition is useful for automatic tagging of photos. These applications and their requirements have generated tremendous interest in face recognition research and development. While the origins of machine face recognition date back 50 years, the general problem of face recognition is incredibly difficult due to large intra-person face variability: pose, illumination, expression, occlusion, and aging. In other words, different face images of the same person acquired at different times and under different imaging conditions can have quite different appearances that are difficult to match by state of the art systems. Hence, the challenge is to design salient feature extractors and robust matchers for face images. For this reason, convolution neural networks, also known as, deep networks, have played a major role in the new generation of face recognition systems. This talk will address a number of ongoing research projects in my laboratory that include (i) face identification at scale (gallery of 80 million faces), (ii) large-scale face clustering (120 million faces), (iii) longitudinal study of face recognition, and (iv) detection of spoof faces.

Who Is There? Face Recognition Applications

部门:

计算机科学系

讲者:

JAIN, Anil

发布日期:

2015

影片长度:

70 分钟

影片类型:

学术论坛

There are two main drivers of face recognition technology: (i) security, namely, access to restricted areas and personal devices, de-duplication of passports and driver licenses and identifying suspects in surveillance videos, and (ii) social media, where face recognition is useful for automatic tagging of photos. These applications and their requirements have generated tremendous interest in face recognition research and development. While the origins of machine face recognition date back 50 years, the general problem of face recognition is incredibly difficult due to large intra-person face variability: pose, illumination, expression, occlusion, and aging. In other words, different face images of the same person acquired at different times and under different imaging conditions can have quite different appearances that are difficult to match by state of the art systems. Hence, the challenge is to design salient feature extractors and robust matchers for face images. For this reason, convolution neural networks, also known as, deep networks, have played a major role in the new generation of face recognition systems. This talk will address a number of ongoing research projects in my laboratory that include (i) face identification at scale (gallery of 80 million faces), (ii) large-scale face clustering (120 million faces), (iii) longitudinal study of face recognition, and (iv) detection of spoof faces.

Full Disclaimer

This video is presented here with the permission of the producers. Any downloading, storage, reproduction, and redistribution are strictly prohibited without the prior permission of the respective producers. Go to Full Disclaimer.