IGCV3: Interleaved Low-Rank Group Convolutions for Efficient Deep Neural Networks

Ke SunMingjie LiDong LiuJingdong Wang

   Papers with code   Abstract  PDF

In this paper, we are interested in building lightweight and efficient convolutional neural networks. Inspired by the success of two design patterns, composition of structured sparse kernels, e.g., interleaved group convolutions (IGC), and composition of low-rank kernels, e.g., bottle-neck modules, we study the combination of such two design patterns, using the composition of structured sparse low-rank kernels, to form a convolutional kernel... (read more)

Benchmarked Models

RANK
MODEL
REPO
CODE RESULT
PAPER RESULT
ε-REPRODUCED
BUILD
1
IGCV3 x1.0
72.1%
--
2
IGCV3 x0.75
68.9%
--
3
IGCV3 x0.5
60.2%
--
4
IGCV3 x0.25
46.3%
--