TensorFlow Tutorial Step 1: Set Up for THETA V
First steps in modifying Tensorflow code to work inside the THETA V
Code for Makoto Shohara’s workshop is available on GitHub.
This is Step 1 of the workflow described in How to Build Tensorflow apps for RICOH THETA.
Note: if you want to run TensorFlow in the THETA first prior to working on the code, read Running Tensorflow on RICOH THETA V
Starting code (tag): WS_START_STEP1
Overview
- We’re going to start at the point the “PPP” error sound stops when you first start the plug-in
- Use the Camera API to modify the behavior
Fixing “PPP” Error Sound
- Build the sample code and run it on the THETA V
- The plug-in on the camera will stop with a “PPP” sound.
- Use Vysor to set the permissions of the plug-in.
- If the camera still stops with a “PPP” error sound, follow the comments in the code “Step 1” to end the error without opening the camera resource
Using Camera API
notificationCameraClose()
sendBroadcast(new Intent(“com.theta360.plugin.ACTION_MAIN_CAMERA_CLOSE”)
free the camera resources that is being used by the main camera application
The camera resource can only open one application
Camera Parameter Settings
parameters.set(“RIC_SHOOTING_MODE”, “RicMoviePreview1024”);
Set to camera operation mode. The default is RicStillCaptureStd
Depending on your requirements, you can change RIC_PROC_STITCHING settings as follows:
- Static stitching (high speed): “RicStaticStitching”
- Dynamic stitching (high precision): “RicDynamicStitchingAuto”
- No stitching (Fish-eye): “RicNonStitching”
Using the Microphone
AudioManager am = (AudioManager) getSystemService(Context.AUDIO_SERVICE);am.setParameters(“RicUseBFormat=false”);
In order to do real time processing, you should use single channel mono audio.
When you use Ambisonic B-format (4-channel) audio, it will be recorded as a WAV file.
Next Step
TensorFlow Tutorial Step 2: Incorporating PluginLibrary
