Open PhD Position
Efficient edge machine vision is critical for realizing many next-generation applications with edge intelligence, like smart cars, robots. Machine vision is an exploding market with a size of 13.23 billion USD in 2021 and will have a compound annual growth rate of 7.7% from 2022 to 2030.
The goal of the project is to build sub-milliwatt activity-driven computing hardware that can support the inference and learning of machine vision tasks at the extreme edge.
The PhD student is expected to work on two sub-projects:
On the software side, you will design model compression methods to reduce the computational and memory cost of state-of-the-art machine vision algorithms for video processing tasks, such as human activity recognition/object tracking. You will also investigate efficient on-chip learning algorithms for machine vision.
On the hardware side, you will develop energy-efficient activity-driven hardware accelerators based on field programmable gate arrays (FPGA) and application-specific integrated circuits (ASICs).
Please submit your application on the TU Delft Job Vacancy Site.