Build and deploy custom object detection model with TensorFlow Lite
summary
Shows how to train a custom object‑detection model with TensorFlow Lite Model Maker and export a quantized .tflite, then drop it into an Android Studio app using the TensorFlow Lite Task Library. Uses an EfficientDet family model (example: EfficientDet‑Lite0), trains on ~100 labeled food images in about 30 minutes, exports salad.tflite, and integrates in a few lines of Kotlin. Measured AP50 is about 39.6% (drops to ~35.6% after TFLite quantization); the app successfully visualizes bounding boxes.
AI-written summary · about
chapters (6)
- 0:00Start
- 5:32Object detection vs. image segmentation
- 6:09Object detection use cases
- 6:45Using a pretrained model
- 7:34Training your own model
- 8:22Step 1: Create a training dataset
related talks
17mDeploy a custom Machine Learning model to mobile
34mTensorFlow Lite
40mWho’s afraid of Machine Learning? First steps with TensorFlow for Android.
27mMachine Learning on the edge using TensorFlow Lite
17mBuild a smart IoT device with TensorFlow Lite and Raspberry Pi
42mCreating Smarter Applications
19mOn-device machine learning: TensorFlow on Android
13mMLOps Pipelines to Deliver Tf.Lite Models to Distributed Compute Pools at the Edge