Deployment bug-fix #1036
Deployment bug-fix #1036
Commits on Jan 27, 2016
Commits on Jan 28, 2016
Commits on May 6, 2016
Commits on May 8, 2016
Commits on May 11, 2016
Commits on May 12, 2016
Commits on May 13, 2016
Commits on Jul 13, 2016
-
-
Merge branch 'devel' into new-solvers
Conflicts: examples/cnn_train.m examples/cnn_train_dag.m
Commits on Sep 17, 2016
Commits on Sep 25, 2016
-
Makefile: bug fix: make sure that specified CUDA libraries are linked…
… instead of the MATLAB default ones
-
-
-
-
-
-
-
-
-
Commits on Sep 26, 2016
Commits on Sep 27, 2016
Commits on Sep 29, 2016
Commits on Sep 30, 2016
Commits on Oct 3, 2016
Commits on Oct 4, 2016
Commits on Oct 5, 2016
Commits on Oct 6, 2016
Commits on Oct 24, 2016
Commits on Nov 8, 2016
Commits on Nov 22, 2016
-
Fixes vlfeat#694, libjpeg not aborted once a libjpeg error found.
Plus added a bit more helpful messages for non-existent images.
-
Fixes once more the issue vlfeat#552.
Added a bit more explanations for this decision to prevent future miscommunication.
-
Commits on Nov 24, 2016
-
Fixed the
vl_imreadjpeg.mhelp string.Did not show fully in MATLAB due to additional newline.
Commits on Nov 30, 2016
-
Seems to work for training the Variational autoencoder. Yet to add unit tests etc.
Commits on Dec 1, 2016
Commits on Dec 2, 2016
Commits on Dec 9, 2016
-
You need to review this changes and run this example. The previous version was giving errors.
-
Changed the transpose so that it doesn't give any error. It displays both top1err and top5err for both BSLN and BNORM
Commits on Dec 18, 2016
Commits on Dec 19, 2016
-
Merge pull request vlfeat#805 from oya163/patch-1
Updates the `cnn_mnis_experiment` launch script fixing the names of the error fields in the `cnn_train` output.
Commits on Jan 18, 2017
-
-
Added a fix for the new version vl_imreadjpeg.
Adds a `cropSize` element from the border value, used in previous versions.
-
Small improvements of cnn_train:
* added epochSize option (number of samples per epoch, allows to adjust the epoch size without reducing training set size) * Stats are not saved in the eval mode * Added title to the diagnostics.
-
-
-
Merged in caffe-dilate-fix (pull request vlfeat#26)
added dilation to caffe Conv import (fixes vlfeat#816)
Commits on Jan 20, 2017
Commits on Feb 8, 2017
Commits on Feb 9, 2017
Commits on Feb 13, 2017
Commits on Feb 15, 2017
-
Merged in smallTypos (pull request vlfeat#23)
minor typo fixes to vl_imreadjpeg docs Approved-by: Karel Lenc Approved-by: Joao Henriques
Commits on Feb 22, 2017
Commits on Mar 16, 2017
Commits on Mar 17, 2017
-
-
Merged in loss-improvements (pull request vlfeat#30)
Loss improvements Approved-by: Joao Henriques <[email protected]>
Commits on Mar 22, 2017
Commits on Mar 23, 2017
-
Removed unnecessary try/catch.
At the end, the original purpose of having a network all the time in a valid state does almost never works as the network fails on rebuild. The rethrow makes it harder to debug issues.
-
Loss.m: number of samples for spatial maps now handled correctly (instead of crashing). vl_nnpdist.m: Handles correctly aggregate for 4D tensors.
-
-
Fix of the numerical derivatives for the vl_nnpdist.
With aggregate -> single scalar, numerically the algorithm is even worse. Decreased the size of the input tensor.
-
Merged in toy-data-example (pull request vlfeat#28)
Minimal training/custom IMDB example with toy data, and tensor visualization function * added cnn_toy_data example, and vl_tshow to visualize tensors * fix vl_tshow for ndims < 4 * decouple toy image generation from IMDB creation, more realistic use-case * vl_tshow: allow setting parent axes and other properties * vl_tshow: replace hacky montage+padding with imagesc+grid, better scaling * moved files to examples/custom_imdb/
-
Merged in vl_argparse-improvements (pull request vlfeat#31)
Improvements to vl_argparse: 'field.subfield' syntax, ignore missing defaults (all optional) Approved-by: Samuel <[email protected]> Approved-by: Andrea Vedaldi <[email protected]>
-
Merged in docFix (pull request vlfeat#25)
fixes dropout explanation updates vl_nndropout doc string to define rate as the probability that a variable is dropped. Approved-by: Andrea Vedaldi <[email protected]>
-
-
Commits on Mar 24, 2017
-
-
Commented out unused bool variable which causes "Unused" warnings.
Originally used only in debug mode. The bilinearsampler_gpu needs cleanup.
-
-
