ResNet 50 - AshokBhat/ml GitHub Wiki

About

  • 50-layer ResNet
  • Trained on more than a million images from the ImageNet database
  • Achieved a top-5 error rate of 3.57% in 2015 which beats human-level performance on ImageNet dataset

Characteristics

  • Depth - 50 layers
  • Input size - 224x224
  • Can classify images into 1000 object categories, such as a keyboard, mouse, pencil, and many animals

Accuracy

Performance

Example code

Graph

See also