Hello,
I have high résolution images which I want rather to tile before applying inference (rather than resize). One solution could be to modify the inference code inside the docker but I am looking to know if either 1) there is a parameter that I can set somewhere to let the inferencer know that it should tile the image before or 2) if someone know about a publicly available solution for this.
Best
In TAO Faster_rcnn, there is not tiled inference yet. We do have similar solution in nvocdr(link).
For public solution, you can take a look at How to Use Tiling During Inference, How to Use SAHI to Detect Small Objects.
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. Thanks
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.