I started a new project in deep learning with caffe. I need to get lots of RGB/Depth/infrared images and some other information from kinect 4 windows and analyze them with caffe. Caffe uses leveldb/lmdb for large scale data import/export. leveldb is installed on my c++ project and its wotking correctly. Now, what is the most efficient way to save images on leveldb? is it enough to convert image to a byte array and put in in leveldb?
Asked
Active
Viewed 735 times
2
-
[this post](http://stackoverflow.com/q/37265977/1714410) might inspire you – Shai May 18 '16 at 06:32
-
By saying 'the most efficient way', what do you mean? It is ok to store image to levelDB, but it depends on the kind of performance you care about, the throughout of write or the latency of read. – sel-fish Jun 10 '16 at 03:39