saihv / darknetx2 Goto Github PK

Thank you so much for this project - batching in inference is very important!
(by the way - what is the speedup you get from this?)

Anyhow, I'm been so far unsuccessful in using this - getting CUDA errors (CUDA Error: invalid device pointer) followed by seg faults. This is by trying to infer with detect_np_batch, giving it a 4-D nparray of an image and assuming batchNum is the batch size).
I'm probably doing something wrong there so I hope you can give me a hint on how to properly use this function.

Thanks!