Yuan Liu

Hi, you may refer to https://docs.opencv.org/4.x/dd/d43/tutorial_py_video_display.html

Hi, you may need to use a blender for this purpose. I render it offline and then compose it. It would involve some complicated coordinate system transformations so I do...

Hi, 是c2w。我这里会统一转换为w2c。 https://github.com/liuyuan-pal/Gen6D/blob/50aa71b2dbb3198772c025d54d0dd191a918982c/dataset/database.py#L534

`data\custom\pandas\test/images_out_smooth/%d-bbox.jpg: No such file or directory` 默认输出是在这个文件夹里面，似乎你的这个文件夹是空的

嗯嗯，我指的是，那个predict.py应该会在这个文件夹里面输出一些图片~这个不是mp4的文件夹

你要不看看是不是漏了其中哪个数据集，看看这个names是不是空的~

Hi, thank you for your interest in our work! 1. For training objects, we use both ref and test in training. We then evaluate on unseen objects, which are not...

Yes, there is a map between the COLMAP id and the frame id. You may check https://github.com/liuyuan-pal/Gen6D/blob/50aa71b2dbb3198772c025d54d0dd191a918982c/dataset/database.py#L173 So you can retrieve the input image.

抱歉，这里确实是平均值，在paper里面少除了一个32的三次方

嗯嗯，是的，也没有开方