Curriculum of Developing Deep Learning Application Course by Intel
-
1. Programming Tools and Foundational Concepts
-
Module 1: Introduction to AI and OpenVINO™ Toolkit
-
1. AI market overview and use cases of Deep Learning
2. Deep Learning development cycle - History, performance, and motivation
3. Challenges in developing and deployment of Deep Learning applications at the Edge and while working with various Hardware Platforms
4. Intel® Distribution of OpenVINO™ toolkit overview, its components and development workflow
5. Creating an AI inference in few lines of code using OpenVINO™ toolkit
6. Simplify and Develop for the Edge with Intel® AI Ecosystem and the various hardware and software offerings
-
Module 2: Setting up the Development environment for AI with OpenVINO™ Toolkit
-
1. AI prototyping with OpenVINO™ toolkit using Intel® DevCloud for the edge
2. Run sample applications and evaluate performance on various Intel® architecture with Intel® Devcloud for the edge
3. OpenVINO™ toolkit installation and setup on a local machine
4. Getting started in 5 minutes with OpenVINO™ toolkit using Jupyter* Notebooks
5. OpenVINO™ toolkit opensource version- Clone, Compile and Contribute
6. Intel® Edge Software Hub, a one-stop resource for optimized software and offerings for key edge use cases
-
Module 3: Expedite development of high-performance deep learning inference applications with Intel® Open Model Zoo
-
1. Introduction to Open Model Zoo and download a pre-trained model for learning or for developing deep learning software
2. Analyze various pre-trained models and public models and select an appropriate model based on the use case
3. Tools overview - Model downloader and Accuracy checker
4. Demos demonstrating various use cases with pre-trained and public models
-
Module 4: Optimization and Quantization of models for better performance
-
1. Introduction to Model Optimizer and understand its significance
2. Learn how model optimizer helps in streamlining the AI application development flow
3. Model conversion overview with Model Optimizer and understanding general and framework-specific conversion parameters
4. Generating Intermediate Representation files using Model Optimizer
5. Convert model with general and framework-specific conversion parameters
6. Understanding how the model optimizer works under the hood
7. Choosing the right Quantization option in Model Optimizer based on Hardware Platform and assessing the benefits and trade-offs of using different Quantization
8. Introduction to low precision optimization with Post-Training Optimization Tool and Understand the advantages of INT8 calibration
9. Learn various methods to boost model performance with Post-Training Optimization Tool
-
Module 5: Inference Engine & integration with deep learning applications
-
1. Understanding various Intel® architecture - CPU, iGPU, VPU to achieve accelerated performance
2. Introducing Intel® Movidius™ Myriad™ X VPU and getting started with Inference on Intel® Neural Compute Stick 2
3. Hardware agnostic inference with write once deploy anywhere approach
4. Creating powerful, scalable, and futureproof AI applications with a streamlined Development Workflow using Inference Engine
5. Performance improvement using SYNC and ASYNC modes of the Inference Engine
6. Efficient utilization of all the available compute resources using Heterogeneous and Multidevice Plug-ins to boost the performance
-
Module 6: Advanced Labs
-
1. Writing job file in Intel® Devcloud for the Edge and inference on multiple nodes
2. OpenVINO™ toolkit beyond vision- Speech, OCR, and NLP
3. Using the benchmark app to estimate the inference performance of your deep learning model on various devices
4. Advanced video analytics- Multiple models in one application
5. Local Project Migration with Intel® DevCloud for the Edge
6. Advanced samples and demos
-
Module 7: Streamline AI Application Development with Deep Learning Workbench
-
1. Introduction to Deep Learning Workbench and identifying the benefits
2. Learn to create optimized AI applications using a web-based graphical environment
3. Explore key concepts and features of Deep Learning Workbench
4. Getting familiar with Deep Learning Workbench workflow
5. Getting started with Deep Learning Workbench and identify the difference between running the Deep Learning Workbench on your local system and in the Intel® DevCloud for the Edge
6. Achieve the goal of optimized AI applications, whether you are new to deep learning or an advanced deep learning developer with Deep Learning Workbench
7. Develop and deploy optimized AI applications with Deep Learning Workbench
-
Module 8: Speed up the solution development journey with Intel® Edge Software Hub
-
1. Accelerate innovation for key edge computing use cases and starting with a powerful foundation for data processing, image processing, and edge AI analytics
2. Selecting the vertical-specific software package, customize the configurations to your needs, and decide on your target hardware for deployment
3. Explore the Reference Implementations to see how businesses are deploying edge intelligence in the real world
4. Downloading a Reference Implementation and Customizing it according to your requirements
5. Simplifying the edge-to-cloud workflow integration with Intel® Edge Software Hub
