迈向自动驾驶之AI物件侦测技术
教师: 張添烜
公务人员MOOC     

2019/03/29 ~ 2019/12/31
3小时/4周 (已经开始)

概要

本课程介绍以深度学习为基础的物件侦测作法,适用于自动驾驶或辅助驾驶设计。将会介绍各种经典的两段式与一段式作法,并探讨如何加速符合即时应用,与如何增加准确率的各种做法,最后将以实例说明实际应用场景所遇到的问题与效果。

This course will introduce object detection approach based on deep learning, which is suitable autonomous driving applications. The topics will cover two stage approaches, single stage approaches and discuss how to accelerate the execution of these deep learning models to meet real time constraints and improve their accuracy. Finally, a real case will be studied to show their strength and weakness.

课程目标

预期学生修完本门课,能深入了解以深度学习为基础的物件侦测背后的原理与其限制,并对实际应用所面临的即时运算与准确率问题,知道如何解决。

After taking this course, students will learn the principle and limitations of the object detection based on deep learning, and know their real time constraints and accuracy problems and their solutions.

授课教师

download.php?id=417&area=teacherfiles

ChangTS.jpg
张添烜博士于2000年于交大电子获得博士学位,于2000~2004任职于创意电子担任副理,于2004年加入交大电子担任教职至今。于2009年至比利时IMEC担任访问学者。张博士曾获中国电机工程学会优秀年轻电机工程师奖,台湾IC设计学会杰出年轻学者奖。他的专长为VLSI 设计,深度学习与讯号处理。

Tian-Sheuan Chang received the B.S., M.S., and Ph.D. degrees in electronic engineering from National Chiao-Tung University (NCTU), Hsinchu, Taiwan, in 1993, 1995, and 1999, respectively.

        From 2000 to 2004, he was a Deputy Manager with Global Unichip Corporation, Hsinchu, Taiwan. In 2004, he joined the Department of Electronics Engineering, NCTU, where he is currently a Professor. In 2009, he was a visiting scholar in IMEC, Belgium. His current research interests include system-on-a-chip design, VLSI signal processing, and computer architecture.

  Dr. Chang has received the Excellent Young Electrical Engineer from Chinese Institute of Electrical Engineering in 2007, and the Outstanding Young Scholar from Taiwan IC Design Society in 2010. He has been actively involved in many international conferences as an organizing committee or technical program committee member.

课程进度表

第1周:Two stage detection 两段式物件侦测

第2周:Single stage object detection 一段式物件侦测

第3周:Fast object detection and small object detection 快速物件侦测与小物件侦测

第4周:Improvement and real case study 效果增进与实例探讨

课程内容

周次

单元主题

第一周

Two stage detection

两段式物件侦测

1-1 Object Detection

1-2 R-CNN

1-3 Fast R-CNN

1-4 Faster R-CNN

1-5 R-FCN

第二周

Single stage object detection

一段式物件侦测

1-6 Single Stage Object Detection

1-7 SSD: Single Shot MultiBox Detector

1-8 YOLO v2 v3

1-9 RetinaNet

第三周

Fast object detection and small object detection

快速物件侦测与小物件侦测

1-10-1 PVANet

1-10-2 Object Detection at 200FPS

1-11 Comparison

1-12 Mask R-CNN

1-13 Small Object Detection

第四周

Improvement and real case study

效果增进与实例探讨

1-14-1 Improvement Over Mask R-CNN and RetinaNet-1

1-14-2 Improvement Over Mask R-CNN and RetinaNet-2

1-15 Uber Event

上课形式

本课程分为 15 个单元,每周配合课程内容提供随堂测验,以帮助学习者快速确认是否了解上课内容,另安排各单元测验用以考核学习成果,考核标准请参见「评分标准说明」。

评分标准

  • 平时测验:各单元测验,共 15 单元,占 100 %

通过标准


课程及格标准:60分满分:100分

先修科目或先备能力

本课程建议具备基本深度学习概念即可,无须太多背景知识,适合所有对深度学习于物件侦测有兴趣的学习者修习。