yu4u
commented
4 years ago Hello, it seems that the code method is to find a model directly from pretrain, and then modify the output category to 101, and then do a simple classification, whether this will not work well. Do you need any special treatment?
from google translate
Although solving age estimation as classification problem works well, in prediction, expectation of estimated age probability is used as final output as proposed in [1].
[1] R. Rothe, R. Timofte, and L. V. Gool, "DEX: Deep EXpectation of apparent age from a single image," in Proc. of ICCV, 2015.
zhanglaplace
您好，看代码方法好像是直接从pretrain找了个模型，然后修改输出类别为101，然后做一个简单的分类，这个是不是效果不会很好。需要做什么特殊的处理吗～。