UROP Proceedings 2020-21

School of Engineering Department of Computer Science and Engineering 129 Machine Learning on Wearable Devices Supervisor: HUI Pan / CSE Student: WU Chi-hsuan / DSCT Course: UROP1100, Fall As wearable devices become more and more handy and gradually increase its popularity, information around us is likely to be recorded and uploaded to the internet more easily in the future. At the same time, people’s consciousness of protecting their own privacy have raised in these years. In this report, we propose a method to protect people’s privacy by cartoonizing the faces on the image and modify specific facial segments. The cartoonization of faces provide deidentification protections of individuals in the image while preserving the user experience and satisfaction when sharing the protected images in online services such as social networks. First, we use face detection model and cartoonization model to get the cartoonized faces. Next, we get the contour of different segments of the faces and crop them into ready-to-use images. We believe distorting facial segments could further protect people’s privacy on the image. Machine Learning on Wearable Devices Supervisor: HUI Pan / CSE Student: DING Ziheng / CPEG WANG Ziyi / SSCI Course: UROP1100, Summer UROP1100, Summer The project ‘Machine Learning on Wearable Devices’ is aiming at exploring interaction between users and programs on wearable displays through hand gesture recognition. It consists of 2 parts: ‘Copy and Paste’ and ‘Hand Pose Estimation by IMUs’. ‘Copy and Paste’ project is mainly about recognising hand gestures by camera to do ‘copy’ or ‘paste’ tasks, without any contact device. ‘Hand Pose Estimation by IMUs’ is mainly about recognising hand gestures by Inertial measurement units(IMU), which serves as an addition when hands are out of the detection range of the camera. Part I is made by WANG Ziyi and Part II is made by DING Ziheng. Parameterized Algorithms for Static Program Analysis Supervisor: KAFSHDAR GOHARSHADY Amir / CSE Student: BIDAIBEK Sanzhar / COMP Course: UROP1100, Spring During this UROP1100 course under the guidance of Professor Amir Goharshady, I successfully studied the introduction into parametrized algorithms as applied in the context of static program analysis, as well as the basics of infinite automata theory, and was assigned and started the work on a research topic of optimizing methods related to HyperLTL Model Checking. I conclude that this was a fruitful experience and I am eager to continue the work in the upcoming semesters.