questions for test

[liyuanyaun]

in test.py, it will concatenate the train images with each test image as the input of the model. and based on experiments, i find the result is different between with and without concatenation. why this happened? could you answer me? thank you very much @dawei6875797

[aco92]

in test.py, it will concatenate the train images with each test image as the input of the model. and based on experiments, i find the result is different between with and without concatenation. why this happened? could you answer me? thank you very much @dawei6875797

Do you know why he concatanate it at all? I am analyzing the code, and seems pretty complicated and unclear logically. For example, if you have test images and you want to make corresponding aged copies which belongs to other age groups, why it has to bi linked with images from train folder?

[liyuanyaun]

in test.py, it will concatenate the train images with each test image as the input of the model. and based on experiments, i find the result is different between with and without concatenation. why this happened? could you answer me? thank you very much @dawei6875797

Do you know why he concatanate it at all? I am analyzing the code, and seems pretty complicated and unclear logically. For example, if you have test images and you want to make corresponding aged copies which belongs to other age groups, why it has to bi linked with images from train folder?

maybe due to its BN(batch normalization)， it must use train mode to test the model.

[aco92]

I don't get it. Could you clarify?