Skip to content

Transfer Learning using ISET & Matlab

David Cardinal edited this page Oct 25, 2022 · 1 revision

The transferLearning directory provides a tutorial on taking a pre-trained image recognition network and "tweaking" it by letting the top layers also learn from your own dataset. This is an industry-wide technique for efficient, application or dataset specific, use -- as it is very expensive to train high-quality models from scratch, and the basic feature recognition process is often very similar across a variety of uses.

Clone this wiki locally