Icecream-blue-sky

How to implement arcface without Partial FC with pytorch?

1

I find that it is forced to use partial fc in arcface_torch, but I just want to use arcface. Is there any offical implement with pytorch? Or need I implement...

In arcface, is there any difference set no_bias=True or False? ```python In recognition/arcface_mxnet/train.py fc7 = mx.sym.FullyConnected(data=nembedding, weight=_weight, no_bias=True, num_hidden=config.num_classes, name='fc7') ```

Why only output the max cls score during inference?

7

I notice that YOLOX only output the max cls score for each anchor point during inference, why ?

Since there is softmax-version focal loss, is there also a softmax version of general focal loss?

Why not use multi-scale?

1

I am curious why multi-scale is not applicable, is it because of the amount of calculation?

Actually, I don't know why most one stage detectors use sigmoid classifier even though COCO is a multi-classification task. Do you have a more reasonable explanation？I'm really confused about it....

Why not limit the regression range for each location of each FPN level?

1

In FCOS, each FPN level is responsible for predicting objects of different sizes. But in DW, each object is predicted by all FPN levels, why? 

Why the params of IResNet50 is larger than ResNet50?

1

IResNet50 vs ResNet50(in object detection): 43.77M vs 33.71M