AI related software projects：
Decorate your background
Make meetings more fun and personal with Decorate your background. Meeting participants can now use generative background effects in Teams to show up their best – even when the space they’re working from isn’t at its best. With Decorate your background, meeting participants can use the power of AI to generate a background that decorates and enhances their real-world room, such as by cleaning up clutter or adding plants to a wall. [2022.9 - Now]
Animated backgrounds
The animated backgrounds feature in Teams meetings allows you to replace an existing background with a dynamic animation for a more immersive virtual environment. It offers various options to enhance meeting experiences with creativity and personalization based on what you want. [2022.3 - Now]
Green Screen
The new feature green screen in Teams meetings improves the sharpness and definition of the virtual background effect around your face, head, ears, and hair. It also allows you to show a prop or other object in your hand to be more visible to other participants in a call. [2021.11 - Now]
Soft focus and Adjust brightness
Soft focus and Adjust brightness for Microsoft Teams. Enhance the video quality when lighting is poor by turning on the Adjust brightness filter. You also can use Soft focus to apply a subtle smoothing effect. [2021.11 - Now]
Deep Learning based ADAS PoC on Intel GO Platform
[A big success as design win] The 1st Object (car, person, truck, traffic lights and speed limitation signs) detection and road segmentation PoC/reference design on Intel Go (Car Branch: Deverton + A10) platform, for which we performed deep learning model training, customization and optimization, with Intel A10 FPGA acceleration. [developed within 2016.9-2018.4]
Open source machine translation plus post-editing
[Used by intel internally] Developed a machine tranlsaion engine based on Moses, with its corresponding web service API, windows client, and web page access. With these capabilities, it can be used to help to reduce cost for Intel's localization process, by using the model called 'MT + Post-editing'. This MT engine is being maintained and used for Intel internal for a long period until now. [2009.1-2010.12]
Vibration based Message Exchanging
[POC] Use vibration to exchange messages and achieve simple control among different smart devices, no need to pair, just put them together. This is an innovation project I ever developed during working at Intel. [2015.1-2015.3]
AI based diagnosis assistance: Pulmonary nodule detection and analysis
[Production project, used in China hospitals] AI based pulmonary nodule detection and analysis [2021.7-2021.11]
AI based diagnosis assistance: Breast molybdenum target detection and analysis
[Production project, used in China hospitals] AI based breast molybdenum target detection and analysis [2021.7-2021.11]
AI based pathological mechanism analysis
[Production project, used in China hospitals] AI based pathological mechanism analysis [2021.7-2021.11]
AI based multiomics (radiomics, CNN, geonomics, and pathologic omics) research platform
[Production project, used in China hospitals] The platform can extract features from radiomics, geonomics, and pathologic omics, then combine them with various ML models for different research purpose. [2021.7-2021.11]
AI based Auto Contouring for OAR(Organs At Risk)s
[Production project] AI based Auto Contouring for OAR(Organs At Risk)s. [2019.6-2021.6]
Face Recognition for Patient (with Masks) ID Auto Confirmation
[Preproduction]Face Recognition for Patient (with Masks) ID Auto Confirmation. [2020.1-2020.6]
4D CT Image Registration
[POC, conference paper] 4D CT Registration. [2020.1-2020.6]
Automatic Gantry Angles Optimization
[POC] Automatic Gantry Angles Optimization. [2020.1-2020.3]
Predictive maintenance
[Exploration] Lead intern to reproduce some predictive maintenance models of some latest papers. [2020.1-2020.4]
Deep learning and GAN based Dose Prediction
[POC] Deep learning and GAN based Dose Prediction. [2020.1-2020.3]
Driver Status detection on Movidius NCS
[POC] The project is develop a driver status detection model on Intel Movidius NCS (Neural Computing Stick) and optimize the performance and accuracy. So far, we reached 20 frames per second on i7 CPU and one NCS unit, and accuracy > 97% on our own testing dataset. [2017.1-2017.3]
Use deep learning and data fusion to detect comprised camera sensors for ADAS and Robot usages
[POC & Patent] This is a research to detect the comprised camera or other sensors using deep learning, for autonomous driving and robot usages, we have tested the algorithm and so far the accuracy has researched 89%. We are also applying a PCT patent for this algorithm for Intel. [2017.2-2017.3]
Deep Learning based Question Answering system
[POC] The project is to research and develop a QA (Question answering) system using deep learning technologies like Word2Vec, LSTM, and etc. [2016.6-2016.9]
Hypertension Risk Predictive Analytics Based on TAP
[Production] The project is to train the Hypertension Risk Model using machine learning cloud supports of Intel’s TAP (Trusted Analytics Platform). I am mostly helping to choose the models and clean the data, as well as machine learning algorithm supports. [2016.1-2016.5]
Media Pal and Voice Interactive Media
[POC] This project is to develop a QT application to let the user be able to interact with the media by using voice. The platform we are using is MeeGo IVI. In this project, I have researched and setup all the necessary environment, and developed a small demo by using QT for the whole process of “audio input -> voice recognition -> Text to Speech”. [2010.12]
Fall Detection
[POC] Work together with team to develop an Android appplication for smart phone (and possibly other smart devices like smart watch etc.) to perform fall detection using embedded acceleration sensor, which can be mostly used by old people for fall detection and real-time alarming. [2018.5-2019.5]
Parking Space Line Detection Algorithm and HW Optimization
[Production] Lead the algorithm team developed the PSLD (Parking Space Line Detection) algorithm and corresponding hardware optimization on TDA2, the algorithm is using traditional computer vision algorithms and runs very fast, it supports outdoor all weather conditions and indoor underground conditions. Addtional sensor fusion with ultrasonic radar detection are also performed to enhance the perception scope and stability. [2018.5-2019.5]
Pedestrian Detection Algorithm and HW Optimization
[Production] Lead the algorithm team developed the PD (Pedestrian Detection) algorithm and corresponding hardware optimization on TDA2, the algorithm is using traditional computer vision algorithms and runs very fast, it supports outdoor all weather conditions and indoor underground conditions. Addtional sensor fusion with ultrasonic radar detection are also performed to enhance the perception scope and stability. [2018.5-2019.5]
Moving Object Detection Algorithm and HW Optimization
[Production] Lead the algorithm team developed the MOD (Moving Object Detection) algorithm and corresponding hardware optimization on TDA2, the algorithm is using traditional computer vision algorithms and runs very fast, it supports outdoor all weather conditions and indoor underground conditions. Addtional sensor fusion with ultrasonic radar detection are also performed to enhance the perception scope and stability. [2018.5-2019.5]
Motion planning for robot arms
[POC] This is a research to apply Guided Policy Search, Imitation learning and ROS to robot arms for smarter motion planning. [2017.4-2017.6]
vSLAM Research for APA and VPA
[POC & Research] vSLAM like DSO and LSD are promising teches for localization and obstacle detection, without the need of knowing what object it actually is. This is researches regarding using this for APA and VPA usages, and corresponding speed optmization on HW. 2 China patents are generated from this research. [2018.5-2019.5]
Edison based smart device design win
[POC & Exploration] This project is develop based on the Edison development kit, to provide basic functions for smart devices like auto login using Bluetooth and wifi, play audio and etc. [2015.4-2015.6]
Ransomware Detection
This project is to detect ransomware using machine learning methods, I was mostly working as a machine learning consultant for this project. [2017.12]
Motion Planning and Control System Design of the Hexapod Robot
[University project] This project is about the analysis and designing of the movement of the Hexapod Robot. In this project I presented a Hexapod Robot Design Principle. The robot fuselage can be made by alloy or stainless steel and designed from the concept of bionics, using Insect Triangle Gait to walk, drive by 12 servos which are controlled by the programming of using Keil C51 to complete the corresponding actions of the six legs, finally achieved to control the robot to move forward or backward, or to turn left or right. [2007.3-2007.6]
Pure software projects：
Localization and I18N tools development and support
[Production, Intel localization automation saved Intel $26.1 million from 2007 to 2010] Maintain and made new tools for the localization and I18N processes (including file, network and software localization operations), most of the tools are developed by using VB, C# or VBA, some are using C++ or perl. [2008.1-2010.12]
China DRM
[Production] The project is to develop the China DRM system based on the former PKI system we developed and the new standard defined. We developed an (should be the first) China DRM Client tested with open source player which can run on Linux, Windows and even Anroid based on the standard.[2013.11-2015.12]
Driver Certificate Signing Web Portal
[Production] The project is to integrate with an internal system and develop a web portal for the driver certificate signing. Before this, this operation is done manually. We automated the signing process, protected it with security methods and developed a web portal for user to do the signing operation, also integrated it with proper user and role management capability to make it much easier and unified for the access application. [2013.3-2013.10]
China Trust Authority
[Production] The project is to design and develop a PKI system to be used and adopted by CCTV/CNTV to manage keys and certificates, which is the basic for a DRM system. My responsibility is to develop the Rights Issuer Authority and its corresponding Disaster Recovery module, also developed a backup and restore tool for the deployment and migration of the database. [2012.4-2013.2]
Android WiFi and NFC Validation and test cases Automation
[Production] The project is to design and ship one Android Tablet based on Intel architecture. Our team is to perform the platform validation on for the tablet, and I am responsible for the validation for the WiFi and NFC (Near Field Communication) domains, and also involved in the testing automation framework design and script programming on Android. [2011.4-2012.4]
Binary Translation Validation Automation
[Production] Binary Translation is used to enable the ARM based Android applications running on IA based platform. In this project, I was responsible for programming the automation script to verify the basic ARM assembly instructions to run successfully and correctly on IA platform via Binary Translation. [2011.1-2011.4]
TMS technical support
[Production] TMS is short for Translation Management System, which is a tool developed by SDL. My main responsibility is to develop scripts based on the programming interface TMS provides to provide some customized reports or functions for Intel TMS users. Sometimes, I also help Intel users to configure the settings of TMS to meet their requirements. [2008.1-2010.12]
GPS Navigation program using EVC
[Intern project] This program is made by using C++. The involved technologies include multithreading, serial port communication, extracting the useful information from GPS data, files operation, multimedia playing and etc. The whole programming and test of this project are finished all by myself. I ever made two versions for this program, one is made by using MFC and the other is programmed only using Windows APIs. [2007.7-2007.11]