projects / platform / upstream / caffe.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: e5cc609) | patch
default backend to lmdb for image conversion and mean computation
authorEvan Shelhamer <shelhamer@imaginarynumber.net>
Sun, 21 Sep 2014 22:20:47 +0000 (15:20 -0700)
committerEvan Shelhamer <shelhamer@imaginarynumber.net>
Sun, 21 Sep 2014 22:35:16 +0000 (15:35 -0700)
commita6693464787b63d805f077961cb39eb96698807f
treecbfa362f5b3ccb877d11474fe7fbf275c7668a61tree | snapshot
parente5cc609138a0bc4ce5177a67cf84952756d11b38commit | diff
default backend to lmdb for image conversion and mean computation

lmdb is 10-15% faster than leveldb although it takes ~1.1x the storage.
This is usually irrelevant in prefetching since both are fast enough,
but more important lmdb allows multiple, concurrent reads for training
and evaluation several models on the same data.
tools/compute_image_mean.cpp diff | blob | history
tools/convert_imageset.cpp diff | blob | history
Domain: Machine Learning / ML Framework;