Pre-trained EfficientNet models (B0-B7) for PyTorch
Tagscomputer science, classification, deep learning, cnn, multiclass classification
Recently new ConvNets architectures have been proposed in “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks” paper. According to the paper, model’s compound scaling starting from a ‘good’ baseline provides an network that achieves state-of-the-art on ImageNet, while being 8.4x smaller and 6.1x faster on inference than the best existing ConvNet.
Pretrained weights from lukemelas/EfficientNet-PyTorch repository