/tvmodels

Pytorch implementation of vision models.

Primary LanguagePythonMIT LicenseMIT

tvmodels

The tvmodels library contains pretrained vision models in pytorch trained on ImageNet. Some of these models are available in torchvision but some are not, so you can load them for here.

Installation

Run the following to install:

pip install tvmodels

Colab

If it shows ModuleNotFoundError on Google-colab use the following:

!git clone https://github.com/rohitgr7/tvmodels.git
import sys
sys.path.append('/content/tvmodels')

Usage

from tvmodels.models import se_resnet50, resnet18

# Load the models
se_res_model = se_resnet50(pretrained=True)
res_model = resnet18(pretrained=True)

Available models

  • ResNet(s)
  • ResNext(s)
  • SEResNet(s)
  • SEResNeXt(s)
  • SENet154
  • EfficientNets