This message was deleted.

Hi Matteo, could you post your

service.py

?

The problem is related to serve option.

If you enable batching (setting

batchable=True

), then

aimv_clf_runner.do_inference.async_run

should take a batch input. So maybe use

imgs = np.expand_dims(img, axis=0)
prediction = await aimv_clf_runner.do_inference.async_run(imgs, measure_times=True)

Then inside

do_inference

remove

unsqueeze

part

Thanks!!