Multiple object image recognition

We build an iOS app prototype to recognize multiple objects in a single still image, leveraging Apple's Vision framework and IBM's Watson technologies for Core ML.

These are the slides from the winning pitch in the category "most demanding implementation" in the 2018 Challenge Days hackathon organized by Die Mobiliar in Bern, Switzerland.

Transcript

1. thuysichanqua.ai MULTIPLE OBJECT IMAGE RECOGNITION
2. Toan & Tanh
4. Applications • Garage Sale (C2C) • Brocki & Retail (B2C) • House Contents Valuation (Insurance)
5. 3 Step Workflow Object Recognition Object Classification Instance Processing Rotation Perspective Correction Masking
6. Technology iOS Vision Framework iOS Core Image Framework IBM Watson for Core ML
7. Demo
8. Performance
9. Privacy
10. Device Identfies and Analyses Updated Model Watson Retrains Model Feedback Hyper-customized Model
11. Conclusion Delightful User Experience Increased Sales User Privacy Protected