Hi, everyone. I am a new in matconvnet, I find the time bottleneck is train data loading in matconvnet.
How to implement data loading and train computation in parallel in matconvnet?
Hi @xhhong - sorry for the slow response. One way to speed up data loading is to make use of the prefetch functionality in vl_imreadjpeg(), which you can see demonstrated here.
Hi, everyone. I am a new in matconvnet, I find the time bottleneck is train data loading in matconvnet. How to implement data loading and train computation in parallel in matconvnet?
thanks!