lawsonxwl

icon company Xidian University icon location Xi'An China

Results 20 comments of lawsonxwl

使用源码config训练CAN-DWAP，在CROHME14测试集上得到的ExpRate比论文中要低1.4%

> 我在测试中得到 52.23% ExpRate。可能是什么原因？ 尝试增加训练epoch数量

使用源码config训练CAN-DWAP，在CROHME14测试集上得到的ExpRate比论文中要低1.4%

> 作者是有自己的数据增强处理，但是这部分代码没有开源，所以达不到很正常 论文中无数据增强数据给的是57% 带上数据增强之后是65.58%

使用源码config训练CAN-DWAP，在CROHME14测试集上得到的ExpRate比论文中要低1.4%

> 使用的是论文的配置 batchsize为4 大了V100装不下

使用源码config训练CAN-DWAP，在CROHME14测试集上得到的ExpRate比论文中要低1.4%

> > @lawsonxwl 我训练了240轮，代码没有任何改变，在2014 test 上只有54.67% ExpRate, 而且把counting loss 和 counting_preds 去掉之后，ExpRate 几乎不变，还是54%左右，和论文里的ablation study 完全对不上，不知道是哪里的问题，你有尝试过去掉counting loss 和 counting_preds 之后训练么 大佬你好 我看你在ICDAR2023的成绩很好 问下是使用的CAN模型吗？ 有没有用啥trick 类似于beam search 和模型集成这种

SDXL checkpoint bad result?

为什么用 test_brushnet_sdxl.py 输出的是inpaint背景的结果？ 我这边用这个代码和图片输出的是inpaint前景的结果。。。

is the traindata model input correct？

> @lawsonxwl you can check it out [#35 (comment)](https://github.com/TencentARC/BrushNet/issues/35#issuecomment-2177491312) thanks for the reply, i have checked that, could you show me your generation result? I have trained SDXL ver brushnet...

is the traindata model input correct？

> @lawsonxwl > > here is some outputs. my input image: ![image](https://private-user-images.githubusercontent.com/61898803/341636449-265e90b3-6ba4-4fe5-b12f-be0db6176c83.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.4cCUU6hrtI_9IUd_DUvrW_ufljSRgcE13zcdc-3ScuU) > > some outpainting results from my burshnet sdxl trained model: > > ![image](https://private-user-images.githubusercontent.com/61898803/341636506-34cce84e-3d1b-4c96-81a7-ebeb6b998688.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.fEQTxhFRUSnn2UaUgk-AFUePPs0W913RAZmnYtBivEs) ![image](https://private-user-images.githubusercontent.com/61898803/341636601-5b93f680-d01e-46f1-8bc5-eb7123f89352.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.GdbOizvSWDStFx6yzKdiIpEUFANSCIx_X0QbU3VmLus) ![image](https://private-user-images.githubusercontent.com/61898803/341636626-dab7925d-da62-454b-8875-957150991009.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.C4TvHR3ujVJnQlpmsf-XYpHtj_vU_3_Zu4o8WvvY3Ig) > >...

is the traindata model input correct？

> @lawsonxwl > > here is some outputs. my input image: ![image](https://private-user-images.githubusercontent.com/61898803/341636449-265e90b3-6ba4-4fe5-b12f-be0db6176c83.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.jJ46U8LJd6_TnJ7PWJeLR_LSIh3pLLW9yy7nw-0rN3k) > > some outpainting results from my burshnet sdxl trained model: > > ![image](https://private-user-images.githubusercontent.com/61898803/341636506-34cce84e-3d1b-4c96-81a7-ebeb6b998688.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.uyYXz8bfGHbICy9fnF45ftWAl_T65BuOm07vrX3r2_U) ![image](https://private-user-images.githubusercontent.com/61898803/341636601-5b93f680-d01e-46f1-8bc5-eb7123f89352.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MTg5NTEyOTgsIm5iZiI6MTcxODk1MDk5OCwicGF0aCI6Ii82MTg5ODgwMy8zNDE2MzY2MDEtNWI5M2Y2ODAtZDAxZS00NmYxLThiYzUtZWI3MTIzZjg5MzUyLnBuZz9YLUFtei1BbGdvcml0aG09QVdTNC1ITUFDLVNIQTI1NiZYLUFtei1DcmVkZW50aWFsPUFLSUFWQ09EWUxTQTUzUFFLNFpBJTJGMjAyNDA2MjElMkZ1cy1lYXN0LTElMkZzMyUyRmF3czRfcmVxdWVzdCZYLUFtei1EYXRlPTIwMjQwNjIxVDA2MjMxOFomWC1BbXotRXhwaXJlcz0zMDAmWC1BbXotU2lnbmF0dXJlPWVlMWNmYzk0MzZiNzc4M2Q0NjBjNjdiYzI3NjMyYmViYjM2MGMzMjk2NWYzMTA0MjEyZDhiYzZiOGQ4MzE5NWUmWC1BbXotU2lnbmVkSGVhZGVycz1ob3N0JmFjdG9yX2lkPTAma2V5X2lkPTAmcmVwb19pZD0wIn0.okOuB4R4ll1XSqPlp6l6d9Nw_3zDzRZzkClIfD3u97U) ![image](https://private-user-images.githubusercontent.com/61898803/341636626-dab7925d-da62-454b-8875-957150991009.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.XKlcrZd5blK4vhA7rUskxQ7Qfu_LUkCk1xzSd2ZCKxE) > >...

is the traindata model input correct？

As u say in the comment: The resulting output of the rle2mask function is mainly a binary mask image where the ones represent the foreground pixels and zeros represent the...

is the traindata model input correct？

thegreatlsx 发自我的iPhone ------------------ Original ------------------ From: zjlinkin ***@***.***> Date: Wed,Jul 3,2024 11:47 AM To: TencentARC/BrushNet ***@***.***> Cc: lawsonxwl ***@***.***>, Mention ***@***.***> Subject: Re: [TencentARC/BrushNet] is the traindata model input correct？...