Margrate

> Hi, > > I have trained OCR on cityscapes many times with different hyperparameters, however, I still cannot get wanted mIoU which is 81.60% (and the highest one I...

You should test with flip and multi scale. My training time is about 20 hours ---Original--- From: ***@***.***> Date: Sun, Sep 12, 2021 20:57 PM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject:...

加我的微信交流  1137096314 ---Original--- From: ***@***.***> Date: Sun, Sep 12, 2021 20:57 PM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject: Re: [HRNet/HRNet-Semantic-Segmentation] performance of pascal-context (#240) Hi there I got mIOU also around...

50.76 is the results with the config FLIP=False paper's results is FLIP=True But I can't get the same Mean_Acc as author by testing the lip model released

train on val(sample 2000 images) mIoU=50.93 testval(10000 images).   mIoU:0.5544,  Pixel_Acc:0.8807, Mean_Acc:0.6617 add My Wechat: 13713916489 加我微信，交流一下HRNet ------------------ 原始邮件 ------------------ 发件人: "HRNet/HRNet-Semantic-Segmentation" ***@***.***>; 发送时间: 2021年9月7日(星期二) 下午5:47 ***@***.***>; ***@***.******@***.***>; 主题: Re: [HRNet/HRNet-Semantic-Segmentation] Do anyone reproduce the...

> The performance on LIP and Pascal Context (COCO stuff and ADE) is stable. But on Cityscapes, the performance is not very stable. Can you got mIoU(59 calsses)=54.0 and mIoU(60...

width(channel) of first branch

> I solved this problem by updating the Pytorch version. Can you get the results mIoU(59 classes)=54.0% and mIoU(60 classes)=48.3%? I got a big performance gap.

> Hi everyone, > > Could you explain for me the meaning of this line > https://github.com/HRNet/HRNet-Semantic-Segmentation/blob/8037a7fb867b58d885b26b1ceb8283a7ee252851/tools/test.py#L82 > > https://github.com/HRNet/HRNet-Semantic-Segmentation/blob/8037a7fb867b58d885b26b1ceb8283a7ee252851/tools/test.py#L83 > > `pretrained_dict = {k[6:]: v for k, v in...

The config multi-scale and flip set as NO both in training and testing or training is YES, testing is NO?