Can you tell me what to do to test on unlabelled dataset?

[sgrrlll]

I have trained and validated on my own dataset and tested on a test dataset using an inference file written by myself but found that the results obtained were not good and the results were good or bad depending on the size of the output image, but if the validation code from the code was used on the test set, the results were found to be good.