Code taken from this tensorflow repo.
This is a camera app that continuously detects persons with and without bicycle helmet (bounding boxes and classes) in the frames seen by your device's back camera, with the option to use EfficientDet Lite 0, EfficientDet Lite1, EfficientDet Lite2, or EfficientDet Lite3 model trained on the Kaggle dataset.
This application should be run on a physical Android device.
-
The Android Studio IDE. This sample has been tested on Android Studio Bumblebee.
-
A physical Android device with a minimum OS version of SDK 24 (Android 7.0 - Nougat) with developer mode enabled. The process of enabling developer mode may vary by device.
-
Open Android Studio. From the Welcome screen, select Open an existing Android Studio project.
-
From the Open File or Project window that appears, navigate to and select the tensorflow-lite/examples/object_detection/android directory. Click OK.
-
If it asks you to do a Gradle Sync, click OK.
-
With your Android device connected to your computer and developer mode enabled, click on the green Run arrow in Android Studio.
Downloading, extraction, and placing the models into the assets folder is managed automatically by the download.gradle file.
You can use this pretrained models to test here.
You can train your own model with this Jupyter Notebook using Google Colaboratory.
View some slides about this app.