Hi,
I am trying to run YoloV2 model with INT8 precision. It requires calibration table to build the engine for INT8 model. I couldn’t find any calibration table for YoloV2. Is there a way to create our own calibration table.
Thanks.
Hi,
I am trying to run YoloV2 model with INT8 precision. It requires calibration table to build the engine for INT8 model. I couldn’t find any calibration table for YoloV2. Is there a way to create our own calibration table.
Thanks.
Hi dumbdog,
You can find more information at [url]NVIDIA Metropolis Documentation
We are making plans for DS next release. Do you have specific requirement for this calibration? we may consider this into roadmap.
Hi windy60j34,
If i may recommend, including some sort of program or sample to create our own calibration table as part of the DS release would be nice. Hope this is helpful
Hi windy60j34,
If i may recommend, including some sort of program or sample to create our own calibration table as part of the DS release would be nice. Hope this is helpful
Thanks for your suggestion. It’s helpful for us to improve deepstream-sdk in future versions.