Even

  基本明白了，跟p,p`在最小二乘中的顺序有关。 还有个疑问：  这里不应该是SE3d么，怎么是SO3d呢？

别附议了，估计太监了...

@husheng12345 Hi, maybe the kitti D1 difference comes from the negative disparity treatment in your implemention

I found the 'segout' head is not equipped with a Sigmoid head, and the training is not going to converge on my observation...

> Traceback (most recent call last): File "/data1/zhongyao/dl/Point-SAM/demo/app.py", line 5, in from point_sam import build_point_sam ModuleNotFoundError: No module named 'point_sam' > > I tried pip install git+https://github.com/zyc00/Point-SAM.git and python setup.py...