-
Hi everybody,
I am currently trying to install the matconvnet-1.0-beta23 version on my Matlab 2015a. My Operating System is Windows 10 Home Edition.
I installed previously the CUDA Toolkit 8.0 o…
-
### Environment info
Operating System: Fedora 25
cuda 8.0 and cuDNN 5.1
TITAN X (Pascal) and Driver Version: 375.26
I installed tensorflow 1.0.0 using pip install under anaconda
https://storage…
-
I want to compute deconvolution in dnn.
But I coudn't find it in "layers.h".
For example, How do I compute shown as below image (The blue panels are sources and the green panels are destinations.)…
-
Hi,
I tried this in several network architectures and by only adding one dropout layer with rate = 0.5, the training become slower and slower until the point it is barely progress.
The weird thing i…
-
Hi,
I was wondering what changes would I need to make to incorporate an Euclidean loss function in place of the standard log loss function.
Would simply passing the outputs of the softmax layer and c…
-
It seems that the latest MatConvNet versions introduced some minor breaks in this code (see issue raised https://github.com/vlfeat/matconvnet/issues/552 ), so I suggest using versions from 1.0-beta18 …
Relja updated
7 years ago
-
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!
-
Hi guys,
I'm using convnets to do pose estimation and try to train the net from scracth. However, even I make the architecture and parameters same, I will get quite different results every time. My d…
-
-
Hi, I've noticed your extractCNN.m picks up FC activations from the 36th layer. As far as I see, 36th layer is FC8 and it is of 2622 dimensions. Shouldn't you pick up FC activations from 34th layer (F…