About ASSD Metric

[STORM-young]

Hello author, I improved the model on the basis of your code, ran experiments on the ISIC-2018 dataset, and calculated the ASSD and HD95 indicators, but the output result of my ASSD average is only about 0.4, but the ASSD in your paper is 7. and 8. and above, did your ASSD do other processing? My IOU has a small boost over yours.Looking forward to your reply

[jcwang123]

Did you normalize the assd scores? In my experiments, I simply utilized the MedPy package to compute these scores.

…

-----原始邮件----- 发件人:STORM-young ***@***.***> 发送时间:2023-08-05 21:57:27 (星期六) 收件人: jcwang123/xboundformer ***@***.***> 抄送: Subscribed ***@***.***> 主题: [jcwang123/xboundformer] About ASSD Metric (Issue #7) Hello author, I improved the model on the basis of your code, ran experiments on the ISIC-2018 dataset, and calculated the ASSD and HD95 indicators, but the output result of my ASSD average is only about 0.4, but the ASSD in your paper is 7. and 8. and above, did your ASSD do other processing? My IOU has a small boost over yours.Looking forward to your reply — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[STORM-young]

I didn't normalize the assd scores,I just do it like this
from medpy.metric.binary import hd, dc, assd, jc
if np.all(output == 0):
assd_ave = 0
hd_ave = 0
else:
assd_ave = assd(output, label)
hd_ave = hd(output, label)
...

[wanghan-code]

Hello author, can you give the detailed code for your standardized ASSD score