Awesome

<img src=https://raw.githubusercontent.com/dmlc/dmlc.github.io/master/img/logo-m/mxnet2.png width=135/> Model Gallery

All models are hosted at http://data.dmlc.ml/mxnet/models/ and licensed under CC0.

CaffeNet

This model is a pretrained model on ILSVRC2012 dataset. This model is able to achieve 54.5% Top-1 Accuracy and 78.3% Top-5 accuracy on ILSVRC2012-Validation Set.

NIN

This model is a pretrained model on ILSVRC2012 dataset. This model is able to achieve 58.8% Top-1 Accuracy and 81.3% Top-5 accuracy on ILSVRC2012-Validation Set.

SqueezeNet

This model is a pretrained model on ILSVRC2012 dataset. This model is able to achieve 55.4% Top-1 Accuracy and 78.8% Top-5 accuracy on ILSVRC2012-Validation Set.

VGG16

This model is a pretrained model on ILSVRC2012 dataset. This model is able to achieve 71.0% Top-1 Accuracy and 89.8% Top-5 accuracy on ILSVRC2012-Validation Set.

VGG19

This model is a pretrained model on ILSVRC2012 dataset. This model is able to achieve 71.0% Top-1 Accuracy and 89.8% Top-5 accuracy on ILSVRC2012-Validation Set.

Inception-BN Network

This model is a pretrained model on ILSVRC2012 dataset. This model is able to achieve 72.5% Top-1 Accuracy and 90.8% Top-5 accuracy on ILSVRC2012-Validation Set.

This model is converted from TensorFlow released pretrained model. By single crop on 299 x 299 image from 384 x 384 image, this model is able to achieve 76.88% Top-1 Accuracy and 93.344% Top-5 Accuracy on ILSVRC2012-Validation Set.

Full ImageNet Network

This model is a pretrained model on full imagenet dataset with 14,197,087 images in 21,841 classes. The model is trained by only random crop and mirror augmentation. This model is able to achieve 37.19% Top-1 accuracy on training data. This model is about 50% more complex than standard Inception-BN Network