Data layer prefetch queue empty during testing process

52 views
Skip to first unread message

Sajid Iqbal

unread,
Sep 21, 2017, 6:06:30 PM9/21/17
to Caffe Users
Hi,
when i try to run test of deeplab code on its supplied dataset, i get following error. Can anyone explain it why this problem occurs.

Best

Sajid Iqbal

unread,
Sep 21, 2017, 6:08:12 PM9/21/17
to Caffe Users


On Thursday, 21 September 2017 18:06:30 UTC-4, Sajid Iqbal wrote:
Hi,
when i try to run test of deeplab code on its supplied dataset, i get following error. Can anyone explain it why this problem occurs.
When i try its test script, i get following message
I0921 17:57:20.418062 31561 caffe.cpp:285] Running for 1449 iterations.
I0921 17:57:20.418121 31561 blocking_queue.cpp:50] Data layer prefetch queue empty
I0921 18:01:57.233201 31561 caffe.cpp:313] Loss: 0


Best

Przemek D

unread,
Sep 22, 2017, 4:50:44 AM9/22/17
to Caffe Users
"Data layer prefetch queue empty" is not an error, it's a standard notification that caffe gives you when some data layer runs to the end of its source and has to rewind to the beginning.

Sajid Iqbal

unread,
Sep 22, 2017, 7:39:03 PM9/22/17
to Caffe Users
Hi, i was working with VGG based Segnet but i never saw this message there. Any how at present i am working with deeplab and using medical images for segmentation. there are five classes in that however when i train it, it converges to one class only. Any suggestion?
Reply all
Reply to author
Forward
0 new messages