-
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…
-
I don't know how to get the detail of the model"**semantic-segmentation-adas-0001**" which is used in my study./I'll be very grateful if you can answer it . Thank you! @openvino-pushbot
-
hi，大家好，非常高兴的告诉大家，百度飞桨论文复现赛第五期已经开始了，本次**论文复现赛**共将有100篇的经典&前沿论文供大家复现，以及新增了**工程落地赛**，详细信息可以参考[AI Studio](https://aistudio.baidu.com/aistudio/competition/detail/126/0/introduction)，大家是否已经迫不及待了呢~
为了帮助…
-
**Will the performance be influenced or not depends on if the blocked format are used. If there is no blocked format used, then there should not be reorder, the performance drop maybe other reasons.**…
-
### Error:
ImportError: libcudart.so.10.2: cannot open shared object file: No such file or directory
### Environment:
Google Colab with GPU turned on
### Installing commands i followed:
#### …
-
Hi,
seeing my last issue being closed: https://github.com/microsoft/DirectML/issues/16
just updated to latest (200626) tensorflow directml to test on "native" Windows:(tensorflow_directml-1.15.3.dev…
-
“模型训练完成，使用下述命令启动评估
python pdseg/eval.py --use_gpu --cfg ./configs/deeplabv3p_xception65_optic.yaml”
这个评估结果也没有发现生成的图片或文件，屏幕上也没有发现。请问在哪里可以看到？
-
配置：
win10
python-3.7.0-amd64
cuda_10.2.89_441.22_win10
cudnn-10.2-windows10-x64-v7.6.5.32
paddlepaddle_gpu-2.0.2-cp37-cp37m-win_amd64
代码：
import paddle
from paddle import nn
from paddle.io …
-
i am a Hunan University bachelor,Can you leave a message to tell me what improvements your project has made compared with icnet? Thank you very much！
-
我看咱们这边实现了这么多网络，有训练好的模型和评测结果么，跟原文做一下对比~