Skip to content
kwyuan edited this page May 13, 2017 · 11 revisions

Intro

Caffe2 is a deep learning framework that provides an easy and straightforward way for you to experiment with deep learning and leverage community contributions of new models and algorithms.

Most important thing is that the performance is good for mobile devices and here we share how Caffe2 rocks on iOS.

Multiple level of developers

We here provide multiple versions for users in case that different people have different levels of requirements.

Check it out here: Versions

Scope

Main purpose of this repo is to share a project build natively based on Caffe2/ OpenCV 2/ Swift3, covering following features eventually (Scope):

  • A demo app able to show how you can wrap Caffe2/OpenCV2 together with Swift3 via Objective-C bridging

  • Feature to show a static image classifier with a pre-trained Squeezenet Caffe2 model

  • Feature to show a real time classifier via Camera with a pre-trained Squeezenet Caffe2 model

  • Feature allowing users to upload their own Photo to test the performance

  • Feature of a dashboard showing the performance of mobile device in real time

  • Feature allowing users to upload their own Caffe2 model and test it

  • Wrap with Metal GPU Acceleration when official code is out

  • *A CocoaPod to allow you guys setup everything with on click (which is friendly for mobile developers but potentially losing some flexibilities of Caffe2/OpenCV2 for Machine Learning Scientists, and therefore, still debatable)

Clone this wiki locally