MobilenetV2 classificator on VOC2010 dataset using augmentation Structure and model files got from https://github.com/xiaochus/MobileNetV2 5 Classes: ['person', 'bicycle', 'car', 'bus', 'motorbike'] Input size 224x224
Results: Total Top-1 Acc: 86.6%
Accuracy distribution by class: Class person acc=0.889922480620155 Class bicycle acc=0.8519108280254777 Class car acc=0.803921568627451 Class bus acc=0.9253968253968254 Class motorbike acc=0.8600953895071543