Towards Efficient U-Nets: A Coupled and Quantized Approach

In this paper, we propose to couple stacked U-Nets for efficient visual landmark localization. The key idea is to globally reuse features of the same semantic meanings across the stacked U-Nets. The feature reuse makes each U-Net light-weighted. Specially, we propose an
$order$
-
$K$
coupling design to trim off long-distance shortcuts, together with an iterative refinement and memory sharing mechanism. To further improve the efficiency, we quantize the parameters, intermediate features, and gradients of the coupled U-Nets to low bit-width numbers. We validate our approach in two tasks: human pose estimation and facial landmark localization. The results show that our approach achieves state-of-the-art localization accuracy but using
$\sim 70\%$
fewer parameters,
$\sim 30\%$
less inference time,
$\sim 98\%$
less model size, and saving
$\sim 75\%$
training memory compared with benchmark localizers.