Advanced MSC Projects for Computer Science Students

MSC projects for 2018-19 for CMSC5720(term1), CMSC5721(term2) To be supervised by Prof. KH Wong MSC projects for 2018-9 v.7.30 1

Title1: English-to-Chinese translation using machine learning The use of deep learning in machine translation has been shown to outperform pervious approaches as reported in the recent literature. In CUHK, we have successfully built an experimental sentence-based English-to-Chinese machine translator using the sequence-to-sequence approach and we would like to improve the system performance in the coming year. The student is expected to study the method of machine learning and the use of the tool Tensor-flow (https://github.com/tensorflow/nmt ) for the project. The expected result is a high performance English-to Chinese machine translation system with superb accuracy. MSC projects for 2018-9 v.7.30 2

Title2: A virtual reality system of mixed real images and virtual graphics The common virtual reality (VR) system using a head mount device (HMD) only allows users to see the virtual display but blocks the normal views of the environment surrounding the user. It may create a hazardous situation if the user needs to move freely during playing due to the fact that the player cannot see the real surroundings. To combat this problem, we propose to add two cameras to the head mount device and allow the user to see the surroundings by feeding the camera images to the display. By mixing real images with the virtual graphics, it also allows many new possibilities of designing virtual reality games. The student is expected to program a head mount device and use OPENCV (https://opencv.org/) to handle the computer vision and graphical displaying tasks. MSC projects for 2018-9 v.7.30 3

Title3: Wearable optical character recognition device for students and tourists We propose to design a wearable device that can perform instant Optical Character Recognition (OCR) on what the user is reading. Assume the user is reading a paper of text which contains some foreign words that he/she doesn t understand. We propose to help the user by designing a handheld system which has a laser pointer and a camera. The user can use the laser pointer to point to the unknown word and the camera feeds the images of the characters to the OCR module to recognize the meaning. The results can be output using synthesized speech to tell the user of the meaning of the word in his/her mother tongue. Such a device is expected to have high commercial value and useful for students and tourists. The student will use an OCR system (e.g. https://github.com/tesseract-ocr/tesseract) and OPENCV (https://opencv.org/) for this work. A previous similar approach has been achieved by us and the demo can be found at https://www.youtube.com/watch?v=j7t7a8maBv0. The improvement this year is to use a laser pointer instead of a finger to point to the word. And we also try to enhance accuracy, usability and flexibility etc. in this project. MSC projects for 2018-9 v.7.30 4

End Q&A MSC projects for 2018-9 v.7.30 5