You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, @skimmy12, the short answer is no. However, it is possible to convert the TF saved model to a TRT model. Following that, the predictor needs to be modified to accommodate TRT APIs which are slightly different. Additionally, I am not sure how much gain in overall processing speed that could achieve, as a significant portion of processing time is currently consumed by various manipulations on media files. It would be useful to benchmark the improvement and any help on that will be welcomed.
Do you use TensorRT, if so, what version?
The text was updated successfully, but these errors were encountered: