PaddlePaddle / Paddle

PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice (『飞桨』核心框架,深度学习&机器学习高性能单机、分布式训练和跨平台部署)
http://www.paddlepaddle.org/
Apache License 2.0
22.08k stars 5.55k forks source link

pdseg实现auxiliary loss #27823

Closed lyn1225 closed 2 years ago

lyn1225 commented 3 years ago

语义分割领域常用auxiliary loss来优化训练,官方提供的pdseg文件loss.py中定义了multi_softmax_with_loss()函数,并且在config.py文件中也定义了cfg.MODEL.MULTI_LOSS_WEIGHT。如果要实现auxiliary loss,是否只要在相应的模型文件中返回logits的元组,并且在config.py文件中设置logits的元组所对应的MULTI_LOSS_WEIGHT就可以了?菜鸡希望大佬的解答。

bjjwwang commented 3 years ago

已收到,我咨询下Seg的同学回复下。

wuyefeilin commented 3 years ago

目前静态图没有提供支持auxiliary loss来优化训练, 里面的multi_softmax_with_loss()函数是因为icnet而设置的。 我们正在往动态图迁移,在动态图版本里是支持任意增加auxiliary设置的,如果你感兴趣的话可以尝试使用我们的动态图版本。

你目前使用的哪一个模型,目前的训练精度是多少。

有问题的话可以提到https://github.com/PaddlePaddle/PaddleSeg repo的issue里方便我们跟踪恢复。

paddle-bot-old[bot] commented 2 years ago

Since you haven\'t replied for more than a year, we have closed this issue/pr. If the problem is not solved or there is a follow-up one, please reopen it at any time and we will continue to follow up. 由于您超过一年未回复,我们将关闭这个issue/pr。 若问题未解决或有后续问题,请随时重新打开,我们会继续跟进。