adnappp / Sfm-python

三维重建算法Structure from Motion(Sfm)的python实现
212 stars 67 forks source link

最后的点云显示左右是颠倒的 #1

Open zhchyang2004 opened 5 years ago

zhchyang2004 commented 5 years ago

好! 我用crazyhorse的几张图像运行了你的代码. 最后出来的点云显示, 左右是颠倒的. 不知道问题出在哪里?

crazyhorse的图像来自于: https://github.com/TarcLO/SimpleSFM/tree/master/data/crazyhorse

caokui1221 commented 4 years ago

嗨喽,我用的和你一样的图像数据,得到的点云很少,根本看不出来是什么。你能得到正确的结果吗?

zhchyang2004 commented 4 years ago

嗨喽,我用的和你一样的图像数据,得到的点云很少,根本看不出来是什么。你能得到正确的结果吗?

你好! 我得到的点云是正常的,只是左右是颠倒的.

BoxueC commented 3 years ago

嗨喽,我用的和你一样的图像数据,得到的点云很少,根本看不出来是什么。你能得到正确的结果吗?

你好! 我得到的点云是正常的,只是左右是颠倒的.

麻烦问一下,您设置的焦距是多少,我得到的点云特别少

zhchyang2004 commented 3 years ago

嗨喽,我用的和你一样的图像数据,得到的点云很少,根本看不出来是什么。你能得到正确的结果吗?

你好! 我得到的点云是正常的,只是左右是颠倒的.

麻烦问一下,您设置的焦距是多少,我得到的点云特别少

K = np.array([[3140.63, 0, 1631.5],[0, 3140.63, 1223.5],[0, 0, 1]])

BoxueC commented 3 years ago

谢谢,这个焦距是怎么得到的,我从照片详细信息里面得到的焦距和这个不一样

zhchyang2004 commented 3 years ago

我不记得当时为什么用这组内参数据了. 你得到的点云少, 是不是因为正常的点云簇被有异常的飞点给缩成一点了?

lss1008 commented 3 years ago

line 102, in init_structure p1, p2 = get_matched_points(key_points_for_all[0], key_points_for_all[1], matches_for_all[0]) IndexError: index 0 is out of bounds for axis 0 with size 0 请问两位大哥,这个错误是为啥呢?

zhchyang2004 commented 3 years ago

line 102, in init_structure p1, p2 = get_matched_points(key_points_for_all[0], key_points_for_all[1], matches_for_all[0]) IndexError: index 0 is out of bounds for axis 0 with size 0 请问两位大哥,这个错误是为啥呢?

get_matched_points里面的输入参数有问题. 你向前反推,看看是哪一步出现了问题.

lala1108 commented 2 years ago

line 102, in init_structure p1, p2 = get_matched_points(key_points_for_all[0], key_points_for_all[1], matches_for_all[0]) IndexError: index 0 is out of bounds for axis 0 with size 0 请问两位大哥,这个错误是为啥呢?

get_matched_points里面的输入参数有问题. 你向前反推,看看是哪一步出现了问题.

您好,我也出现了这个问题,我也没想明白是哪一个参数有问题

lala1108 commented 2 years ago

line 102, in init_structure p1, p2 = get_matched_points(key_points_for_all[0], key_points_for_all[1], matches_for_all[0]) IndexError: index 0 is out of bounds for axis 0 with size 0 请问两位大哥,这个错误是为啥呢?

您好,请问您解决这个问题了吗

opreationlife commented 1 year ago

图片资源路径配置后无法运行呀,这对图片有啥要求吗

vemacular commented 1 year ago

line 102, in init_structure p1, p2 = get_matched_points(key_points_for_all[0], key_points_for_all[1], matches_for_all[0]) IndexError: index 0 is out of bounds for axis 0 with size 0 请问两位大哥,这个错误是为啥呢?

可能因为keypoint_for_all是空的,可以看看前面特征提取的地方,在最后的return那加上dtype=object

bailanwang123 commented 1 year ago

setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (11,) + inhomogeneous part.报这个错误该怎么处理

bailanwang123 commented 1 year ago

可以问您一些问题吗

vemacular commented 1 year ago

setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (11,) + inhomogeneous part.报这个错误该怎么处理

我之前好像也是这个报错,试一试改成np.array(keypoint_allimage,dtype=object)这种形式看看

ddjiayou commented 4 months ago

设置具有序列的数组元素。请求的数组在 1 维后具有不均匀的形状。检测到的形状是(11,)+不均匀部分。

我之前好像也是这个报错,试一试改成np.array(keypoint_allimage,dtype=object)这种形式看看

你好,请问在哪里修改这个呀

ddjiayou commented 4 months ago

嗨喽,我用的和你一样的图像数据,得到的点云很少,根本看不出来是什么。你能得到正确的结果吗?

你好!我得到的点云是正常的,只是左右是颠倒的. 你好, setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (7,) + inhomogeneous part.这个错误怎样改呀,我用的跟你的数据一样