RESEARCH EXPERIENCE
Motion Detection with Epilepsy-related. (2025- )
This project develops a sensor-based system for detecting and classifying human motion in indoor environments,
with a focus on epilepsy-related pathology. By integrating IMU sensors (accelerometer and gyroscope) and ultrasonic sensing.
The goal is to enable real-time health monitoring for elderly individuals or patients prone to epileptic seizures, apply machine learning models like CNNs and RNNs to classify motion types.
The project emphasizes robustness in noisy environments and practical deployment in smart home settings.
Micro-Attention Modeling in Short Video. (2024- )
This project models user attention in short-form videos by generating micro-attention heatmaps from multimodal content cues.
Without relying on eye-tracking data, the system uses motion analysis, facial expression detection, audio energy tracking, and scene transitions to infer attention peaks over time.
Techniques include optical flow extraction, RMS energy computation, and facial landmark tracking, combined with 1D CNNs and Transformer-based temporal models trained using weak supervision and contrastive learning.
The framework supports applications in video pacing, educational content design, and privacy-preserving engagement analysis.
Short Video wastage. (2023-2025)
The project aims to understand the causes of short video data wastage in terms of viewing time and buffering.
Analyzed the reasons for the waste of different short video servers at the pre-load level.
Designed an algorithm to reduce the short video wastage from pre-loading and buffer level.
Measurement Study across Short Video Services. (2021)
This work provided a comprehensive comparison of four popular
short video services. In particular, the work focus on exploring content characteristics and evaluating
the video quality across resolutions for each service.
Congestion Prediction in WiFi Video Streaming. (2020)
This research aims to detect the network changes to deliver a better rate adaptation mechanism for
adaptive video streaming. This research proposes an algorithm, Congestion Prediction-DASH (CP-
DASH), which is designed to prevent stalling and maximize video quality.
Jump Activities Recognition. (2018)
The purpose of this research is to extract the motion information when jumping, such as jumping
up and down. Through Recurrent Neural Network modeling (RNN), the movement characteristics of the jump
are separated, and extracted the jump information to recognize the specific activities.
Activity Recognition based on radar (Walabot) sensor. (2017)
The research proposes an ambient radar sensor-based solution to recognize the activities
that humans normally perform in indoor environments.
The radar sensors detect abnormal and instantaneous motion situations by high-frequency radio signal.
Tracing location based on Ultrasonic sensor. (2016)
This research work designs and develops a noninvasive distance-based user localization and tracking solution, DiLT, for smart systems.
DiLT consists of mechanical ultrasonic beam-forming designs to track the motion of the user.