I am using the camera preview along with an face analyzer (mlkit face detection). I noticed that the dimensions of the ImageProxy returned from the analyze function are not the same as the resolution i ve set in the config. That is causing issues with the bounding box for the FaceGraphic. Is there something i am missing with setting the resolution in the analysis config?
Asked
Active
Viewed 334 times
0
makis.k
- 432
- 6
- 23
-
The output dimension is on a best-effort basis, because the camera hardware may not support the dimension configured by the app. What is the specific issue are you are running into? – Xi 张熹 Jan 30 '21 at 17:54
-
Did you manage to get it work? – Balasubramanian Jun 17 '21 at 07:33
-
@Balasubramanian unfortunately no. Let me know if you find any solution on this one – makis.k Jun 21 '21 at 22:37
-
I can able to make the changes in ImageAnalysis and PreviewConfig. Both have an independent config to control the target resolution. Maybe you can try with this – Balasubramanian Jun 22 '21 at 06:16
-
@Balasubramanian I ve tried adjusting the different configs. Can you share you code to have a look how you've done it please? – makis.k Jun 22 '21 at 08:48
-
what `res` are you expecting from imageProxy? bcz it has a limit of `1080` – Balasubramanian Jun 22 '21 at 10:23
-
https://www.youtube.com/watch?v=kuv8uK-5CLY&t=774s – Balasubramanian Jun 22 '21 at 10:47