tentone / monodepth

Python ROS depth estimation from RGB image based on code from the paper "High Quality Monocular Depth Estimation via Transfer Learning"

Home Page:https://tentone.github.io/monodepth/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

GPU utilization is very low(26%), GPU‘s memory is almost full

studycv opened this issue · comments

When I run this project, the depth map will change after the object in the camera changes for about 8 seconds. The release frequency of the depth image is about 3hz, and the GPU's memory is almost full, but the GPU utilization rate is very low, close to 26%.The GPU is an Nvidia RTX 2060.how to solve this problem?

Hello

This is the expected behavior these models use quite a lot of memory, so they required a hight memory GPU to leverage the full computing power.

Thanks a lot!