open-cd
open-cd copied to clipboard
A Change Detection Repo Standing on the Shoulders of Giants
数据集
作者你好,我将SNUNet的CDD数据集换成了levir数据集,其余的参数都没动,跑了100轮,为什么f1分数才69呢,忘解答
I am seeing the project detects binary change maps for sattelite images. Do you think it can be used for indoor scene changes as well? Like this use-case: 
test
关于test.py运行报错如何解决 
(openmmlab) trest@~/open-cd$ python tools/train.py configs/changeformer/changeformer_mit-b0_256x256_40k_levircd.py --work-dir work-dirs/temp/ 10/20 14:03:38 - mmengine - INFO - ------------------------------------------------------------ System environment: sys.platform: linux Python: 3.8.18 (default, Sep 11 2023, 13:40:15) [GCC 11.2.0] CUDA available:...
您好李老师,FC-EF、FC-Siam-Conc、FC-Siam-Diff在S2Looking数据集上四个指标均为0,其他方法没有问题。你在此处optimizer是如何设置的呀?
李老师您好,我看您的评价指标直接用的mmseg自带的方法。但是在方法中recall的计算公式和acc的一样。您是不是直接忽视了acc这个指标。
RT In my opinion, bounding box detection is simpler and lighter than semantic segmentation
Regarding the generalization issue of the change detection model, I have read some articles but none of them mentioned it. I don’t know if you know anything here. For example,...
作者您好,我在变化检测利用多头分割,做推理的时候,发现两处有问题,我只改了一处,opencd_inferencer.py文件157行 添加for transform in ('MultiImgLoadAnnotations', ‘MultiImgMultiAnnLoadAnnotations,’ 'MultiImgLoadDepthAnnotation')即可, 第二处是siamencoder_multidecoder.py 文件第253行由于推理是不存在semantic_label,data_sample.gt_sem_seg_from是没有的,还有就是data_sample.gt_sem_seg_to也是没有这个属性的,您看这个怎么改,给个建议,谢谢!