MongoDB World 2019: Gaining ML Insight with Google Vision API and MongoDB
1 like274 views
The document discusses the integration of Google Vision API and MongoDB on Google Cloud, showcasing its applications across various industries like healthcare, retail, and media. It emphasizes the ease of use for extracting insights from images without requiring extensive coding and highlights features such as object detection and OCR across multiple languages. The document also addresses security features, offers a demonstration for users, and outlines next steps for engaging with Google Cloud services.
![19
[End of Slides]](https://image.slidesharecdn.com/madisonday2900amofdraftpresentationmongodb2019-gainingmlinsightwithgooglevisionapiandmongodb-190628145426/85/MongoDB-World-2019-Gaining-ML-Insight-with-Google-Vision-API-and-MongoDB-19-320.jpg)
MongoDB World 2019: Gaining ML Insight with Google Vision API and MongoDB
- 1. Gaining ML insight with Google Vision API and MongoDB on Google Cloud
- 2. Used across flagship products: Google is an AI company Uniqueprojectdirectories Time
- 3. Bringing state-of-the-art AI to the enterprise Retail Financial Services Manufacturing Healthcare & Life Sciences Government Media & Entertainment Technology Energy GamingMarketing
- 4. AI building blocks Sight Language AutoML Video Intelligence AutoML Translation Conversation Structured Data Vision API including Vision Product Search Natural Language API Dialogflow Enterprise Edition AutoML Tables AutoML Vision for cloud + edge models AutoML Natural Language Cloud Text-to-Speech Recommendation AI Video Intelligence API Translation API Cloud Speech-to-Text Cloud Inference API
- 5. Two types of AI building blocks API Pre-trained ML models AutoML Custom ML models Leverage Google’s predefined dataset to automatically detect a vast number of objects, landmarks, logos, etc. No model training required Train your own custom model with labels you define with an easy-to-use graphical interface No coding required
- 6. Vision API Detect popular places and landmarks Classify content with predefined labels OCR support for 50+ languages Detect brands and product logos Identify products from your catalog Identify image properties (colors, etc.) Get hints for best image cropping Detect faces and emotions Moderate explicit content Find similar images on the web Detect objects and retrieve coordinates Extract printed and handwritten text
- 7. Gives journalists a new way to search, access, and analyze millions of historic photos NYT digitized more than a century of perishable photographs and other materials. With the Vision API, Times reporters can now easily search millions of high-res scans to enhance their reporting with even more visual storytelling. Bringing historic content to life Preserves a priceless chronicle of more than 100 years of events that have shaped our world MEDIA & ENTERTAINMENT
- 8. A small team empowers 38 media brands MEDIA & ENTERTAINMENT Reduced cost and the need to rely on outside vendors Enabling marketing, ad, sales teams and more to take full advantage of all content With just 7 people, CBS Interactive is using Video, Natural Language, and Vision APIs to serve 38 digital media brands with content discovery and recommendation solutions.
- 9. Box is using Vision API to help their customers manage and gain insights from their image files, and speed up image-centric processes and workflows. TECHNOLOGY Bringing image recognition and OCR to cloud content management Improved extensive content management for customers in every industry Intelligent structure for 30 billion files managed with powerful capabilities Image source: https://www.box.com/skills
- 10. Use-case ● TBD
- 12. ● end to end security defaults that cannot be disabled: ○ always-on authentication ○ network isolation for dedicated clusters ○ TLS / SSL ○ encryption for data at rest ○ granular role-based access controls ● supports multi-region clusters ● sharding option for high-throughput ● managed backups ● Free Tier available on Google Cloud Platform
- 13. Demo We will demonstrate how easy it is to use the Google Vision API to gain additional insights from a batch of photos that have no prior metadata attached. By using this workflow, we will be able to quickly build a descriptive metadata database that can be leveraged for a variety of business use-cases. ● TBD
- 14. Demo
- 16. Can I try this out myself?... Find all code and steps for this demo here: Or visit cloud.google.com/community and search for “MongoDB” https://cloud.google.com/community/tutorials/mongodb-atlas-appengineflex-nodejs-app
- 17. Next steps... 01 Visit us at our booth Chat with our team, learn more about GCP + Atlas 02 Sign up for a new GCP account and get credits Get a 12-month, $300 credit free trial when you sign up for GCP with a new account. 03 Create a new MongoDB Atlas cluster on GCP Create a free MongoDB Atlas database on GCP
- 18. 18 Thank you
- 21. Document Understanding AI You have a goldmine of documents that is hard, expensive or impossible to tap into. Document Understanding AI lets you easily and cost-effectively extract valuable insights from your documents.
- 22. Process documents & extract insights automatically See what’s there 01 Make it useful 03 Understand it 02
- 23. Detecting objects and understanding language within documents
- 24. Document Understanding AI: How it works
- 25. Image Search: How it works
- 26. Ensures privacy & compliance Engage customers in new and exciting ways Enable shoppers to find products simply by sharing a photo Help reduce friction with product search and purchase Empower retail sales associates with information Detect multiple products in one image Additional use cases Enhance recommendations for similar and complementary products Analyze style trends and competitive pricing Vision Product Search
- 27. Vision Product Search: How it works Customer uploads image