Introduction to object detection with Vertex AI and ML Kit
summary
Shows a concrete pipeline: use Vert.x AI and ML Kit to train an edge model, export as TensorFlow Lite, bundle it into an Android app and run live camera object detection. Team trains on three SKUs, ~100 photos each, waits about 1h40 for a training run, and ships an app that correctly logs pesto, tomato sauce and spaghetti most of the time. Key failures: need negative examples, varied lighting/backgrounds, training time and poor scaling beyond a few dozen products.
AI-written summary · about
chapters (6)
- 0:00Introduction to object detection
- 3:37Smart shopping cart project
- 11:30Integrating ML Kit in Android
- 16:53Live demo of the app
- 17:56Key learnings from the project
- 19:40Closing and invitation
related talks
40mVertex AI: Pipelines for your MLOps workflows
29mFirebase MLKit — AI in Android Town!!
30mMachine Learning in Android Apps with ML Kit
12mML Kit: Turnkey APIs to use on-device ML in mobile apps
10mMachine Learning for mobile
29mFrom TensorFlow to ML Kit: power your Android application with ML
39mCustomize Your App with MLKit
32mReal-Time Object Detection with CoreML