search

saki4510t / UVCCamera

library and sample to access to UVC web camera on non-rooted Android device
3.02k stars 1.21k forks source link

Pixel formats of the IFrameCallback #82

Open thisum opened 8 years ago

thisum commented 8 years ago

I'm trying to integrate the output of the UVCCamera to google vision API(android). My plan is analyse one frame at a time, in order to image analysis on those images(BarCode reading) So the plan is to use IFrameCallback and analyse frames(only 3 frames per second). According to google API, I can use only NV16, NV21, or YV12 formats. So to use the image analysing method, I'm using the following method:

public Frame.Builder setImageData (ByteBuffer data, int width, int height, int format)

Hence I used mUVCCamera.setFrameCallback(mIFrameCallback, UVCCamera.PIXEL_FORMAT_NV21) to setup the camera and ImageFormat.NV21 to create the frame object. But I'm not getting any results.

Could you please let me know, have I used the correct method to get a ByteBuffer in the required format? Or do I need to do some more changes in camera parameters? And also whether the usage of IFrameCallback ok?

saki4510t commented 8 years ago

Hi,

I don't use Google Vision API yet and this is just my assuming.

Do you call ByteBuffer#clear before setImageData within IFrameCallback? IFrameCallback#onFrame is called from native library and argument ByteBuffer is direct ByteBuffer and some fields do not set correctly. And you also needs to call one of setPreviewDisplay or setPreviewTexture with setFrameCallback.

You can also use setBitmap with How to create Bitmap using IFrameCallback

saki