SAM-CD icon indicating copy to clipboard operation
SAM-CD copied to clipboard

Pytorch code of the SAM-CD

Results 4 SAM-CD issues
Sort by recently updated
recently updated
newest added

想试图用多卡训练提速,修改参数后会报以下错误 …… …… 900/1024 images loaded. 950/1024 images loaded. 1000/1024 images loaded. (256, 256, 3) 1024 val images loaded. **Segmentation fault (core dumped)** 修改了 run_encoder模块后还是报同样错误 ![sam-cd1](https://github.com/ggsDing/SAM-CD/assets/56397588/e2ee9299-681d-4552-a40d-45263d86f5af)

![Image](https://github.com/user-attachments/assets/c9dd08dd-8a0d-4a77-aec3-955abd7a6657)

Hi, I recently utilized the SAM-CD algorithm in my own project and came up with a solution to editing the source files in the Ultralytics project. The solution is simply...

作者你好,为什么我训练得到结果无法达到论文里的结果?特别是pre和recall这两个指标。在WHU-CD数据集训练后的评估结果如下图所示(方框里的是文章结果,红色字体是我的结果): 我的训练参数沿用的是作者给定的默认值,并且没有使用随机裁剪操作,只用了随机翻转。请问还有什么参数是我没有调整好的吗? 还有就是代码中的学习率迭代公式与论文中的不一样,应该以哪个为准? 恳请作者解答,谢谢!