I am following the sample from TFT Github sample.
From line 490-512 are for config parameters.
Is --batchSize for inference batch size for TRT engine during inference?
So how to calibrate with multiple images?
How to set calibration batch size during calibration to create trt.engine from uff model?
Yes, --batchSize value is used during the TRT engine creation and TRT inference. Please refer below link.
Calibration batch size is currently set to 4. If you want you can try changing the value.