MAP question

[mel0615]

sorry to bother. I have trained my own dataset on your code with Efficientnet as a backbone, but when calculating MAP with 'mode=MAP', it seemed that the input_size of testdataset didnot match the trained model and got an error.
I checked the code and founded that the images for TRAIN had been resized in the function 'get_random_data()' after reading, while images for TEST hadnt. I cant tell why.
so I wonder if you could help me to solve this problem plz, or just tell me whether it is necessary to call the function 'get_random_data()' in utils.py when I parse the testdataset.