Nvinfer topk

(1) the topk in Gst-nvinfer Class-attributes Group Supported Keys (here ) is depreciated. is there any alternative of this ???

(2) for a Image to Text model, for the same license plate, I sometimes get multiple text. for eg If the license plate is
A123 then I might get, A123 and A12S. how can I select the best out of all ?

There is no update from you for a period, assuming this is not an issue anymore.
Hence we are closing this topic. If need further support, please open a new one.

topK is just a part of the NMS algorithm, and it only works with detection models. What makes you think it is depreciated?

Which model are you using? I think you are asking for the postprocessing of your model(network), but we don’t know anything of your model(s), how can we provide suggestions? Can you ask the model owner to provide the solution?

We have a license plate identification sample with TAO models. NVIDIA-AI-IOT/deepstream_lpr_app: Sample app code for LPR deployment on DeepStream (github.com). You may refer to it.