Service	Build	Release
CognitivePipeline.API
CognitivePipeline.Functions
FaceExplorer
ClientSDK
ShopManager App	TBD	TBD

A 360 Intelligent Experiences workshop around using AI + Mobile Apps to deliver advance intelligent that help auditor and store managers to evaluate Retail market product stocking compliance

The workshop aims to educate and inform data scientists and application development teams on the importance of having Data-Driven Intelligent Apps.

The workshop will give you insights and hands-on sessions on how to leverage innovative scalable Azure AI Services and Cloud Backend Services for your applications that will help you deliver an unrivaled customer experience.

Attendees will also learn how Microsoft's Cognitive Services and Advance AI Techniques can be used in conjunction with App Development Tools to scope, design, and execute the next generation of intelligent services.

Workshop Requirements

Technical Requirements

1. Azure Subscription
2. Visual Studio Code (Mac or PC)
3. Optional - Visual Studio (for PC or for Mac) with Xamarin & latest updates installed - DOWNLOAD
4. Postman - DWONLOAD
5. Azure Storage Explorer DOWNLOAD
6. Python with Jupyter notebooks (required for bonus tracks) DOWNLOAD

NOTE: Visit the Setup section of this workshop to get further details on how to setup various environment requirements.

Target Audience

1. Data Scientists
2. Application Developers
3. Development Leads/AI Leads
4. System Architects

NOTE: We highly recommend you setup and configure your system before attending one of the guided sessions. Although we’ve allowed an hour in the morning to assist in trouble shooting configurations, we won’t have time to do a fresh installation.

Objectives

1. Demonstrate the AI platform capabilities and options
2. Show case the ease of infusing AI in the dev cycle for building AI-models, new apps and innovating existing ones
3. Demonstrate unmatched time to value through leveraging Microsoft AI Platform
4. Deliver omnichannel intelligent apps with a consistent and outstanding end user experience

Guided Walkthrough

You can find a walkthrough guide for the project(s) that outlines how you too can build and deploy your own version of the app, Postman and backend services.

NOTE: We highly recommend you setup and configure your system before attending one of the guided sessions. Although we’ve allowed an hour in the morning to assist in trouble shooting configurations, we won’t have time to do a fresh installation.

1. Setup

2. Architecture Options

3. Cognitive Services - OCR

4. Cognitive Services - Face

5. Cognitive Services - Custom Vision

6. Cloud Backend & Cognitive Services

7. Shop Manager App

8. Bonus Content

Proposed Agenda

Topic	Duration
Registration and Setup	60 mins
Intelligent Experiences Today and Tomorrow	40 mins
Intelligent Retail - Shop Manager Overview	20 mins
Break
Solution White-boarding	30 mins
Computer Vision – Options	30 mins
Computer Vision - Cognitive Services Setup	60 mins
Lunch
Computer Vision - Deployment	60 mins
Bring Intelligence to Mobile	60 mins
Break
Conclusion and Next Steps	30 mins
Q and A	30 mins
Bonus Workshop Material (Take home tasks)
Bonus 1 - Policy Recognition with Resnet & Transfer Learning	-
Bonus 2 - Policy Detection with Faster RCNN	-
Bonus 3 - Mobile DevOps for accelerated innovation	-

Backround

Contoso Shop Manager

Overview

Contoso Market has a huge distribution network of supermarket chains across over fourteen countries. Each of these distributors is required to arrange drinks on their stands according to standardized policies. Each policy describes what shelf a given drink should be on and in what order it should be stocked.​ There are huge costs associated with "routine" audit activities to enforce these policies. Contoso Market wanted to develop a system in which an auditor or store manager could take a picture and be told immediately whether the shelf was stocked correctly, like in the image below.​

Investigation

AI Options for Classification/Object Detection

During our scoping, we investigated a couple of approaches to image classification including:

• Custom Vision Service
• Transfer Learning using CNTK ResNet
• Object Detection with CNTK Fast-RCNN.

App Development

• Highly available Mobile Backend as a Service (MBaaS)
• Apps that targets: iOS, Android and Windows
• Apps must be native to leverage 100% of device capabilities
• Cost efficiency is key (for both development and maintenance)

DevOps

• Establish Continuous Integration & Continuous Delivery for app delivery

Implementation

About the project

I tried to make sure I cover all aspects and best practices while building this workshop, but all included architecture, code, documentation and any other artifact represent my personal opinion only. Think of it as a suggestion of how a one way things can work.

Keep in mind that this is a work-in-progress, I will continue to contribute to it when I can.

All constructive feedback is welcomed :)

Support

You can always create issue, suggest an update through PR or direct message me on Twitter.

Authors

Mohamed Saif	Petra Korica	Sherif El Mahdi
GitHub	GitHub	GitHub
Twitter	Twitter	Twitter
Blog	Blog	Blog

Contributors

• Sethu Raman
• Steve Michelotti for the great Face-API-Explorer web app (GitHub)

Inspired by the Microsoft App Innovation Workshop

This project documentation style is inspired by and makes use of assets and guides from the Microsoft App Innovation Workshop created by Mike James & Robin-Manuel Thiel.