郑启航

official code use `selective_adacolor` superpixel method training 15999 steps results：    

|test images|official code|pytorch version| |--|--|--| |||| |||| |||| |||| |||| |||| |||| |||| |||| |||| ||||

Hi. I add new weights in google drive, you can find in the readme. I also upload tensorflow version weights named `whitebox-tf.zip`.

I found the strange color caused by `guided filter`, but now I didn't find a better method to solve it.

ok. I will try if time permits

@GustavoStahl thanks, I missing the test_code ε value is not equal to train_code ε. But I don't think the *color transfer algorithm* is needed. For this model, it needs to...

mobilenetv1 0.75-yolov3 with 2 detect layers, .h5 file about 13mb, 65 [email protected] on voc . But this repo cannot be achieved, because the optimization part is done for others. I've...

@EunseongBoo you need to use a callback to monitor accuracy. usually, I training 75 epoch got the best model.

应该是数据集类别数量和训练时指定的类别数量不同导致的。你可以将自定义数据集做成voc的格式，这样防止自定义数据集时写代码处理出错的可能性。

检查你生成的数据集label，看看有没有大于6的值。 发自我的iPhone ------------------ 原始邮件 ------------------ 发件人: Kearney ***@***.***> 发送时间: 2022年3月16日 23:40 收件人: zhen8838/K210_Yolo_framework ***@***.***> 抄送: 郑启航 ***@***.***>, Comment ***@***.***> 主题: 回复：[zhen8838/K210_Yolo_framework] 自定义数据集，成功导出anchor.npy和img_ann.npy，然后在训练时却发生如下错误 (#20) 蹲一下后续。我用公开的数据（http://faculty.neu.edu.cn/songkechen/zh_CN/zdylm/263270/list/index.htm），六类数据 修改了参数 CLSNUM=6 ，莫非是我其它参数没改？ make anchors DATASET=voc...