ShuffleNet v2

GPTKB entity

Statements (28)
Predicate Object
gptkbp:instanceOf gptkb:convolutional_neural_network
gptkbp:application image classification
object detection
mobile vision tasks
gptkbp:author gptkb:Xiangyu_Zhang
gptkb:Jian_Sun
Hai-Tao Zheng
Ningning Ma
gptkbp:bench gptkb:ImageNet
gptkbp:designedFor efficient computation on mobile devices
gptkbp:developedBy gptkb:Megvii_(Face++)
gptkbp:digestSize 224x224
gptkbp:feature lightweight design
channel shuffle
channel split
improved speed-accuracy tradeoff
https://www.w3.org/2000/01/rdf-schema#label ShuffleNet v2
gptkbp:introducedIn 2018
gptkbp:notableFor real-time inference
low computational cost
high efficiency on ARM devices
gptkbp:notablePublication ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
gptkbp:openSource yes
gptkbp:predecessor gptkb:ShuffleNet
gptkbp:publishedIn ECCV 2018
gptkbp:repository https://github.com/megvii-model/ShuffleNet-Series
gptkbp:bfsParent gptkb:ShuffleNet
gptkbp:bfsLayer 7