PhD Studentship in WiFi-Vision Cross-modality Learning for Human Activity Recognition

Human activity recognition (HAR) is a compelling topic in the fields of ubiquitous computing, with numerous applications including human behaviour understanding, smart healthcare, human-computer-interaction (HCI), and more. Among various sensing technologies, Radio Frequency (RF) signals such as WiFi can be applied in a less intrusive manner, offering significant potential in smart home environments. However, due to multipath propagation, WiFi signals often contain a high level of noise from physical environments that need to be suppressed before developing HAR models.

In this PhD project, the student will investigate the following three main directions:

1. Designing effective WiFi feature templates (e.g., angle of arrival, Doppler frequency shift) to capture human behaviour information while suppressing noise caused by various factors such as multipath effect.
2. Developing state-of-the-art cross-modality learning algorithms that can leverage other potential sensing modalities (e.g., 3D skeleton extracted from videos) to provide supervision information for representation learning.
3. Developing practical WiFi(-vision) based HAR systems that can be generalised to different environments in real-world scenarios. The student can begin with existing public datasets (e.g., the MM-FI dataset), but will later need to collect their own dataset that includes real-world challenges (with multiple subjects, in less controlled environments). Additional new RF sensing modalities (such as mmWave) can also be collected and incorporated into modelling.

Please contact Dr Yu Guan (Yu.Guan@warwick.ac.uk) in the first instance.

Funding Details

Home fees paid and an annual stipend at UKRI rate for 3.5 years. EU/International applicants are welcome to apply but will be required to cover the difference between home and EU/international fees.

Advert information