getBatch() was not called when doing int8 calibration

[le8888e]

Hello,
I tried to quantize an onnx model by int8 calibration. I used class BatchStream and Int8EntropyCalibrator2 and tried calibration data format of .batch and .ppm. In both cases, getBatchSize(), readCalibrationCache() and writeCalibrationCache() was called but getBatch() was not. So I can't get a right calibration table.
Could you please tell me what is probably going wrong? Thank you.

[rmccorm4]

Hi @le8888e ,

I have a detailed INT8 calibration sample here: https://github.com/rmccorm4/tensorrt-utils/blob/master/classification/imagenet/README.md

and the calibrator class is here: https://github.com/rmccorm4/tensorrt-utils/blob/master/classification/imagenet/ImagenetCalibrator.py

Please take a look through these as a reference.