Always get the same output tensor when running TFLite and Tesseract

[iamjosh]

Hi Community,

I am working on a project which is related to license plate detection and OCR.
Only using TFLite to detect the license plate from the video device is no problem.
But combining Tesseract to recognize license plate text is failed.
It always get the same output tensor from TFLite interpreter.

It should get different output from different frames of the video device.
But the output is always the same as the first output tensor.

Run python code on IMX8MP EVK with Yocto LF5.10.35_2.0.0 imx-image-full.
What could be the problem?
Thanks

[Bio_TICFSL]

Hello iamjosh,

The BSP don't include tesseract you must be sure that it compiled for arm64, however if you received the same tensor output form the error log mean that you don't change the video, try to add the video to your program.

Regards