DiCENet: Dimension-wise Convolutions for Efficient Networks

8 Jun 2019Sachin MehtaHannaneh HajishirziMohammad Rastegari

We introduce a novel and generic convolutional unit, DiCE unit, that is built using dimension-wise convolutions and dimension-wise fusion. The dimension-wise convolutions apply light-weight convolutional filtering across each dimension of the input tensor while dimension-wise fusion efficiently combines these dimension-wise representations; allowing the DiCE unit to efficiently encode spatial and channel-wise information contained in the input tensor... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Semantic Segmentation Cityscapes test DiCENet Mean IoU (class) 63.4% # 44
Image Classification ImageNet DiCENet Top 1 Accuracy 75.1% # 100
Object Detection PASCAL VOC 2007 DiCENet MAP 65.2% # 21