search

WangYueFt / dcp

346 stars 90 forks source link

code speed problem #6

Closed zhangshabao closed 4 years ago

zhangshabao commented 5 years ago

I tested the code with my data. The running time was 4s, the computer was 1070, and the video memory was 8 gigabytes.What is wrong?Excluding the read data time, the network runs in 2. 2 seconds

WangYueFt commented 5 years ago

I tested the code with my data. The running time was 4s, the computer was 1070, and the video memory was 8 gigabytes.What is wrong?Excluding the read data time, the network runs in 2. 2 seconds

Hi,

If you're using your own data, the problem can be: 1. your point clouds are big; 2. data loading seems slow (it takes 1.8s?).

Best, Yue

zhangshabao commented 5 years ago

I tested the code with my data. The running time was 4s, the computer was 1070, and the video memory was 8 gigabytes.What is wrong?Excluding the read data time, the network runs in 2. 2 seconds

Hi,

If you're using your own data, the problem can be: 1. your point clouds are big; 2. data loading seems slow (it takes 1.8s?).

Best, Yue

emmmm The number of point clouds is about 1w, and the speed of reading point clouds accounts for 1.8s.I use TXT file to read。 There have a good way to read?

cbenitez81 commented 4 years ago

One way to do it, it's maybe to precompute the neighbors using a kd-tree, and modify the database to have this at hand for each epoch. knn complexity it's n^2, so if your cloud is big, the implementation in python can take a while. Open3d has an efficient implementation