An Android application utilising the OpenCV and Tensorflow frameworks to develop a mobile platform that can perform an estimation of age & gender of a subject in a given image.
- Capture Image is used to capture a new image with the devices camera.
- On capturing new image the application will begin face detection, only 1 face can be present in order to perform an estimation.
- Import Image is used to import a new image from the devices gallery.
- Estimation can only be performed on 1 face, if multiple faces are detected in a single image then the user must select a face presented to the user in order to extract it.
- All results will displayed in a separate UI instance.
- General
- Change application logo to updated minimal background icon.
-
Replace Toast messages with Snackbar messages for neater UI alerts. - Add option to open previous estimations.
- Main Menu
- Remove DNN Test button from main menu.
- Capture Image
- Add front facing camera functionality.
-
Add graphic to UI alerting user to rotate screen for better image capture. - Add function that prompts user that they infact confirm to use captured image for estimation.
- Import Image
-
Restyle native gallery intent to theme of application.
-
- Multiple Faces
- Fix bug that removes fragment from view when device is rotated.
- Results
- Fix bug that crashes application when device is rotated or placed in background then resumed.
- Restyle results from model output.
- Add option to save image and/or estimation results to phone.
- Testing
- Add unit tests to test code.
- Add tests to test UI elements and functionality.
- Add tests to automatically test age/gender classifiers. (100 images)
- Test functionality on numerous screensizes, I.e. various mobile screen sizes, tablets.