Visual Search with Machine Learning on SAP Cloud Platform

Increase productivity and decrease manual work while maintaining optimal management of inventory

Overview

In a manufacturing like industry, identifying missing or broken items is an unavoidable part of regular service processes, where employees and customers need suggestions for replacements fast. It becomes crucial for companies to quickly provide the right set of resources, backed by a powerful analysis on the existing inventory options to optimize the cost and fulfill the business requirements. Usually, the identification of products is the starting point and integral part of business processes like ordering, service fulfilment and support spanning multiple participants, departments and IT-systems.

With the Visual Search based on ML Foundation services in the Cloud, the effort to identify items in catalogs can be minimized and associated business processes accelerated. The ability to combine the Visual Search with state-of-the-art user interfaces like SAP Fiori on one hand and integrate with SAP and non-SAP backend systems via SAP Cloud Platform on the other hand is an additional boon in solving the above problem and offering a consumer grade user experience.

The solution diagram above illustrates a basic architectural pattern for for building a computer vision enabled application like the Visual Search using ML Foundation services on SAP Cloud Platform.

Download the blueprint 
 

Bill of Material - SAP Cloud Platform Components for Licensing Considerations 

The following Bill of Material is for reference purposes only and represents an example of the SAP Cloud Platform services and components required for this use case. Please consult your SAP Account Executive regarding your specific licensing needs.

Sales Item

Function

Licensing metrics

Example

SAP Leonardo Machine Learning Foundation

ML Foundation services are exposed in a bundle as one large service in SAP Cloud Platform - Cloud Foundry. Services include readily consumable pre-trained models, as well as customizable models like Customizable Object Detection, Pre-Trained Similarity Scoring and more. Licensing is based on consumption.

Overall costs are a result of
· Service consumption
· Node type

Consumption is calculated based on node hours. For pretrained services, the runtime of the API calls is summed (~1 second per call). For customized services, the uptime of the service is measured.

Services can run on different node types (Starter, Basic etc.) with different capabilities (CPU, GPU, RAM etc.)

You can start multiple nodes in parallel.

Pretrained Service (e.g. Image Classifier):

60.000 API calls/month = 17 node hours

Customized Service
(e.g. Image Feature Extraction):

Uptime 24/7 per month = 720 node hours

The node type determines a factor the node hours are multiplied with:

Basic: Factor 1
Standard: Factor 1,5
Etc.

Find more information on pricing here: ML Foundation Pricing Calculator

SAP Cloud Platform Application Runtime

Develop cloud-native applications in different programming languages and run them on SAP Cloud Platform Cloud Foundry environment.

Per GB of Memory/Month

1 Instance with 8 GB/Month for Production

PostgreSQL on SAP Cloud Platform

Consume an object-relational database with PostgreSQL on SAP Cloud Platform.

Per GB of Memory/Month

1 Instance with 8 GB/Month for Production

Customers can use the SAP Cloud Platform pricing estimator  to calculate the required investment for a project. Scale up or down on services as required.

Members and partners of SAP PartnerEdge* can evaluate the development of an application for this use case – most development licensing is covered by the packs offered by SAP’s partner licensing services. Click here for details.

*excluding open ecosystem basic.