MS Thesis Defense
Tuesday, March 2, 2021
Virtual Defense via Zoom
Title: 3M-Pose: Multi-resolution, Multi-path and Multi-output neural architecture search
for bottom-up
pose prediction
Abstract: Human pose estimation is a challenging computer vision task and often hinges on carefully handcrafted
architectures. This paper aims to be
the first to apply Neural
Architectural Search (NAS) to automatically design a bottom-up, one-stage human pose estimation model with
significantly lower computational costs and smaller
model size than existing bottom-up
approaches. Our framework dubbed 3M-Pose co-searches and co-trains with the novel
building block of Early Escape
Layers (EELs), producing native modular architectures that are optimized to support dynamic
inference for even lower average
computational cost. To flexibly explore
the fine-grained spectrum
between the performance and computational budget,
we propose Dynamic
Ensemble Gumbel Softmax
(Dyn-EGS), a novel approach to sample micro and macro search spaces
by allowing varying numbers of operators and inputs to be individually selected for each cell. We additionally enforce a computational constraint with a student-teacher
guidance to avoid the trivial search collapse
caused by the pursuit of lightweight models. Experiments demonstrate 3M-Pose to find models of drastically superior speed and efficiency compared
to existing works, reducing computational costs by up to 93% and parameter size by up to 75% at
the cost of minor loss in performance
Biography: Duc is an aspiring graduate researcher hailing from Vietnam.
He has specialized in the application of Machine
Learning in Computer Vision tasks. He is currently pursuing a Master's degree
in Computer Science at Texas A&M
and later a Ph.D. degree in Electrical andoang Computer Engineering at the University Texas of Austin.
No comments:
Post a Comment