-
"alpha: A scalar for focal loss alpha hyper-parameter. If positive samples number > negtive samples number, alpha < 0.5 and vice versa."
your function:
` L=-labels*(1-alpha)*((1-y_pred)*…
-
https://github.com/lyuwenyu/RT-DETR/blob/9f107dab37c402962a4632570e63ec83c30e1ecc/rtdetr_pytorch/src/zoo/rtdetr/rtdetr_criterion.py#L135
在rtdetr_criterion.py的第135行,分类损失被除以num_boxes而不是num_queries:
…
-
In the [HungarianMatcher](https://github.com/IDEA-Research/detrex/blob/main/detrex/modeling/matcher/matcher.py#:~:text=%5B%3A%2C%20tgt_ids%5D-,elif%20self.cost_class_type%20%3D%3D%20%22focal_loss_cost…
-
1. What is the effect of using counters_per_class and focal_loss together and separately?
2. What is the effect of using focal_loss in all [yolo] layers and certain layers(only 1 or 2)?
3. How can w…
-
-
@kuangliu Could you please tell me why you use log_softmax to compute the focal loss instead of the sigmod layer mentioned in the paper?Or I made a mistake in understanding?
-
Just like YOLO or SSD, how to replace original loss function with focal loss in Caffe? Tanks for your answer,this question puzzle me ..
-
-
Perhaps unupdated code for cpu part since it takes gamma[0], alpha[0] while corresponding cuda part takes single values.
-
Hello! First of all thanks for your implementation. You're really awesome.
My question is how to merger the focal loss with SSD architecture as I'm know working on SSD for my project.
1. Is it c…