MobileNetV2: Inverted Residuals and Linear Bottlenecks

In this paper we describe a new mobile architecture, MobileNetV2, that improves the state of the art performance of mobile models on multiple tasks and benchmarks as well as across a spectrum of different model sizes. We also describe efficient ways of applying these mobile models to object detection in a novel framework we call SSDLite... (read more)

PDF Abstract CVPR 2018 PDF CVPR 2018 Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Image Classification ImageNet MobileNetV2 (1.4) Top 1 Accuracy 74.7% # 132
Retinal OCT Disease Classification OCT2017 MobileNet-v2 Acc 99.4 # 4
Sensitivity 99.4 # 3
Retinal OCT Disease Classification Srinivasan2014 MobileNet-v2 Acc 97.46 # 4

Results from Other Papers


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK SOURCE PAPER COMPARE
Person Re-Identification DukeMTMC-reID MobileNetV2 [sandler2018mobilenetv2] Rank-1 71.05 # 42
MAP 50.45 # 47
Person Re-Identification MSMT17 MobileNetV2 [sandler2018mobilenetv2] Rank-1 42.53 # 11
mAP 18.62 # 11

Methods used in the Paper