search

ErdanC / Tooth-and-alveolar-bone-segmentation-from-CBCT

205 stars 30 forks source link

How to set the parameters for centroid clustering? #20

Open Mr-Yellow opened 3 months ago

Mr-Yellow commented 3 months ago

I have trained all modules. When I tried to inference a CBCT, some errors arised in the function "cen_cluster" in "cnt_skl_dect.py". Specifically, there were no elements satisfied "voting_map>20" (line 42). The highest score produced by my model is 11. I think the settings on line 42 and line 59 will affect the number of clusters.

How should I set these to obtain the clustering center for each tooth?

6868king commented 1 month ago

我已经训练了所有模块。当我尝试推断 CBCT 时，“cnt_skl_dect.py”中的函数“cen_cluster”出现了一些错误。具体而言，没有满足“voting_map>20”（第42行）的要素。我的模型产生的最高分是 11。我认为第 42 行和第 59 行的设置会影响集群的数量。

我应该如何设置这些以获得每颗牙齿的聚类中心？

Hello, I would like to ask you a question. When training the second stage of the network, which is multitasking learning, it is necessary to have label images for boundaries and tooth root markers. However, I did not see any code for data processing in this part. How did you handle it? Can you share the code?