Caffe2: Getting Started
Caffe2 is a deep learning framework that was open-sourced by Facebook in April 2017. Caffe2 has been explicitly built for large-scale production deployment and for use in a constrained resource environment such as mobile devices.
What you'll learn
Caffe2 is an open-source deep learning framework and competitor to frameworks such as TensorFlow, Apache MXNet and PyTorch. It's focus is on efficiency and works well with constrained environments such as on mobile devices. In this course, Caffe2: Getting Started, you'll learn the fundamentals of building neural nets and working with Caffe2, get introduced to the Caffe2 Model Zoo and see how you can import models from PyTorch to Caffe2 using ONNX. First, you'll discover the basic building blocks of Caffe2, blobs and workspaces, nets and operators, and put those together to build neural networks to perform tasks such as regression and classification. Then, you'll get introduced to common image pre-processing techniques and the Caffe2 Model Zoo which offers a wide variety of pre-trained models for common use cases. Next, you'll focus on interoperability between the PyTorch deep learning framework and Caffe2 using ONNX, an open source framework for exporting models from one framework to another. Last, you'll use ONNX to move a super-resolution model from PyTorch to Caffe2. By the end of this course, you should be comfortable building and executing neural networks using Caffe2, using pre-trained models for common tasks and using ONNX to move from one framework to another.
Table of contents
- Version Check 0m
- Module Overview 2m
- Prequisites and Course Outline 2m
- Neurons and Neural Networks 5m
- Introducing Caffe2 6m
- Demo: Installing Caffe2 5m
- Blobs and Workspaces 3m
- Demo: Blobs, Workspaces, and Operators 7m
- Demo: Activation Functions as Operators 4m
- Demo: Building and Executing Nets 6m
- Demo: The Operators Catalog 1m
- Training a Neural Network Using Gradient Descent 4m
- Demo: Linear Regression Exploring the Boston Housing Dataset 4m
- Demo: Linear Regression Using Model Helper and Brew 6m
- Demo: Classification Exploring the Iris Dataset 2m
- Demo: Classification Building and Training The Neural Network 6m
- Module Overview 1m
- Convolution, Pooling and CNN Architectures 4m
- Image Preprocessing Techniques 4m
- Demo: Image Preprocessing Techniques 6m
- Demo: Resizing, Rescaling, and Cropping Images 5m
- Transfer Learning 3m
- The Squeeze Net Pre-trained Model 3m
- Demo: Downloading a Pretrained Model and Setting up the Image Processing Pipeline 6m
- Demo: Preprocessing Images for Prediction 3m
- Demo: Predict Using a Pre-trained Model 4m