Page 59 - Undergraduate Research Opportunities Program Proceedings 2017-18

School of Engineering
Department of Civil & Environmental Engineering

like human beings. This report provides an overview about 1. An introduction to the machine learning 2. How to
implement sensor and transmit data on site.

Advanced Landslide Identifications, Classifications, and Prevention in Hong Kong Using Deep Learning

Supervisor: WANG Yu-Hsing / CIVL
Student: CHAUDHRY Mukund / PHYS Course: UROP2100, Fall

With the rise of services like Google Street View, which allows the user to view images of streets from around the world,
privacy becomes an important issue. Two major sources of concern for privacy are faces of pedestrians and car number
plates. This study attempts to automate the process of blurring pedestrian faces and vehicle number plates. The
problem is treated as a computer vision problem. The image is processed by two modules, the first one being a deep
neural network trained for semantic segmentation, and the second being a combination of several computationally
lightweight computer vision techniques.

Supervisor: WANG Yu-Hsing / CIVL
Student: CHENG Log G / COGBM Course: UROP1100, Summer

In this UROP 1100 project, I am responsible for creating a Raspberry Pi based Real-time, Lidar Odometry and Mapping
(LOAM) device to be use in the future for creating a 3d mapping landscape in Hong Kong. Using a Velodyne's new
PUCK™ (VLP-16) sensor connecting with a raspberry pi 3+ that was installed with ROS and existing Simultaneous
localization and mapping (SLAM) algorithms, we created point cloud of the scanned area by moving the device around.

Supervisor: WANG Yu-Hsing / CIVL
Student: YIU Ka Yan / CIVL Course: UROP1100, Fall

I have taken this Civil Engineering based UROP project which focuses on technology development in light of increased
adaptation of such advanced technologies into all professions, Civil Engineering being one of the major industries to be
heavily affected and possibly revolutionized. Having learnt the basics of Python programming has enabled me to
understand the technology that is being used in the lab more. Whilst I am beginning to learn the hardware aspect of the
project, for example learning the components used in the project and using Printed Circuit Board drawing software such
as Eagle and Altium Designer to design the PCB boards, I have learnt most about the practical applications of such
technology in the real world.

Designing Reliable, Scalable Software Systems for Internet of Things (IoT)

Supervisor: WANG Yu-Hsing / CIVL
Student: TAN Tun Jian / CPEG Course: UROP2100, Fall

This work is a continuing from the work from last summer and including an additional task which is also related to deep
learning. In this work, I am recapping and polishing my knowledge on deep learning concepts to find better solutions in
classify the aerial photograph. Apart from that, I helped in labelling the images to create datasets for training a deep
learning model which is able to identify the pedestrian and number plate of the car such as image blurring. Nowadays, in
the applications such as Google Map, users are allowed to view the photo of the streets which had been taken. Before
publishing them to the public, private information has to be removed.

49