Ipython Notebook showing how to do transfer learning with a resnet50 network


Transfer Learning for Deep Neural Networks with Keras

  • What is the purpose of this POC: In an Jupyter Notebook show, how to do transfer learning with Resnet50.

Link to DEMO or 30s VIDEO of the project


  • What is the benefit for the customer: A customer might use a trained model to detect animals in Zoo, or butterflies or whatever the context is usefull.
  • What is the benefit for the business: A business owner might quite fast train a DNN to recognize things that are of interest.


  • How much MD did you use to build it: 4 MD incl. writing.


  • Which data is it based on: Imagenet Data that was used to train the original neural networks. But the actual training data just need to be roughly 10000 pics, per category.

Used Technologies

  • Which technologies were used: Keras, Resnet50 architecture, Deep Learning


  • Which things could be build with this POC if you had more time: Build the app that the blog post suggest. So have a model that might distinguish animals in a zoo.