zihaomu

Hi @ukoehler, thanks for your report. The error was due to the reduce node cannot process the constant input. I'm considering whether to support this case.

> @zihaomu Anything else is blocking this PR from merge? > > Update: we need to merge with benchmark results. Could you run benchmarking with this model? @zihaomu This PR...

We can directly support ppocr-DB v2 and v3 at new API.

> I do not see issues with the patch, but other platforms should be handled in CMake correctly. Hi @asmorkalov. Thanks for your code review. As discussed with Vadim before,...

Link:https://github.com/opencv/opencv/issues/21078

Hi @leleliu008, I got a workaround. You can directly modifiy the xnnpack source code which is inside the tflite build folder.  The main idea is to bypass the error...

Hi @the-star-sea, any update?

Hi @ukoehler, solving this problem requires full dynamic shape support, which conflicts with the existing engine, and we will fix this problem in the new inference engine of `5.x`.

Yes. Dynamic input support is a very important feature, but currently all memory allocation of the engine is based on fixed memory. And this is why this kind of issue...

This problem will occur when the version of opencv-python is higher than `4.5.5`. This solution works for me: https://stackoverflow.com/a/75054982/14419009. It works well with `opencv-python+PyCharm`.