platform/upstream/caffeonacl.git
10 years agoFix from loss-generalization: accidentally removed mid-Forward return
Jeff Donahue [Wed, 13 Aug 2014 21:54:59 +0000 (14:54 -0700)]
Fix from loss-generalization: accidentally removed mid-Forward return
from PowerLayer (caused bad performance for trivial PowerLayer cases...)

10 years agoMerge pull request #686 from jeffdonahue/loss-generalization
Jeff Donahue [Wed, 13 Aug 2014 20:47:44 +0000 (13:47 -0700)]
Merge pull request #686 from jeffdonahue/loss-generalization

Loss generalization

10 years agoStore loss coefficients in layer; use for prettier training output.
Jeff Donahue [Sun, 13 Jul 2014 23:56:42 +0000 (16:56 -0700)]
Store loss coefficients in layer; use for prettier training output.

10 years agoAdd ACCURACY layer and softmax_error output to lenet_consolidated_solver
Jeff Donahue [Sat, 12 Jul 2014 06:22:21 +0000 (23:22 -0700)]
Add ACCURACY layer and softmax_error output to lenet_consolidated_solver
example.

10 years agoAlso display outputs in the train net. (Otherwise, why have them?)
Jeff Donahue [Sat, 12 Jul 2014 06:06:41 +0000 (23:06 -0700)]
Also display outputs in the train net.  (Otherwise, why have them?)

10 years agoDisallow in-place computation in SPLIT layer -- has strange effects in
Jeff Donahue [Sat, 12 Jul 2014 05:52:02 +0000 (22:52 -0700)]
Disallow in-place computation in SPLIT layer -- has strange effects in
backward pass when input into a loss.

10 years agoAccuracyLayer only dies when necessary.
Jeff Donahue [Sat, 12 Jul 2014 00:21:27 +0000 (17:21 -0700)]
AccuracyLayer only dies when necessary.

10 years agoNet::Init can determine that layers don't need backward if they are not
Jeff Donahue [Sat, 12 Jul 2014 00:19:17 +0000 (17:19 -0700)]
Net::Init can determine that layers don't need backward if they are not
used to compute the loss.

10 years agoMake multiple losses work by inserting split layers and add some tests for it.
Jeff Donahue [Fri, 11 Jul 2014 10:17:53 +0000 (03:17 -0700)]
Make multiple losses work by inserting split layers and add some tests for it.
Test that we can call backward with an ACCURACY layer.  This currently fails,
but should be possible now that we explicitly associate a loss weight with
each top blob.

10 years agoGeneralize loss by allowing any top blob to be used as a loss in which
Jeff Donahue [Fri, 11 Jul 2014 08:55:17 +0000 (01:55 -0700)]
Generalize loss by allowing any top blob to be used as a loss in which
its elements are summed with a scalar coefficient.

Forward for layers no longer returns a loss; instead all loss layers must have
top blobs.  Existing loss layers are given a top blob automatically by
Net::Init, with an associated top_loss_weight of 1 (set in
LossLayer::FurtherSetUp).  Due to the increased amount of common SetUp logic,
the SetUp interface is modified such that all subclasses should normally
override FurtherSetUp only, which is called by SetUp.

10 years agoAdd net tests for loss_weight.
Jeff Donahue [Fri, 11 Jul 2014 08:48:36 +0000 (01:48 -0700)]
Add net tests for loss_weight.

Check that the loss and gradients throughout the net are appropriately scaled
for a few loss_weight values, assuming a default weight of 1 in the loss layer
only.  Also modify test_gradient_check_util to associate a loss of 2 rather
than 1 with the top blob, so that loss layer tests fail if they don't scale
their diffs.

10 years agoAdd loss_weight to proto, specifying coefficients for each top blob
Jeff Donahue [Fri, 11 Jul 2014 08:44:49 +0000 (01:44 -0700)]
Add loss_weight to proto, specifying coefficients for each top blob
in the objective function.

10 years ago[docs] update docs generation for notebook metadata
Evan Shelhamer [Wed, 13 Aug 2014 17:42:30 +0000 (10:42 -0700)]
[docs] update docs generation for notebook metadata

10 years agoMerge pull request #921 from shelhamer/notebook-update
Evan Shelhamer [Wed, 13 Aug 2014 17:33:53 +0000 (10:33 -0700)]
Merge pull request #921 from shelhamer/notebook-update

Fix plotting and metadata in notebook examples

10 years ago[example] change notebook name metadata to avoid conflict
Evan Shelhamer [Wed, 13 Aug 2014 16:27:56 +0000 (09:27 -0700)]
[example] change notebook name metadata to avoid conflict

see https://github.com/ipython/ipython/issues/5686

10 years ago[example] fix plt commands in detection
Evan Shelhamer [Wed, 13 Aug 2014 16:22:42 +0000 (09:22 -0700)]
[example] fix plt commands in detection

10 years agouse plt namespace for imshow in filter_visualization.ipynb
Clemens Korner [Wed, 13 Aug 2014 10:05:30 +0000 (12:05 +0200)]
use plt namespace for imshow in filter_visualization.ipynb

10 years agoMerge pull request #914 from ashafaei/euclidean-loss-fix
Yangqing Jia [Wed, 13 Aug 2014 05:27:43 +0000 (22:27 -0700)]
Merge pull request #914 from ashafaei/euclidean-loss-fix

Fixed the GPU implementation of EuclideanLoss!

10 years agoFixed the GPU implementation of EuclideanLoss to report the loss to the top layer
Alireza Shafaei [Tue, 12 Aug 2014 20:54:51 +0000 (13:54 -0700)]
Fixed the GPU implementation of EuclideanLoss to report the loss to the top layer

10 years agoMerge pull request #846 from qipeng/mvn-layer
Jeff Donahue [Tue, 12 Aug 2014 20:04:02 +0000 (13:04 -0700)]
Merge pull request #846 from qipeng/mvn-layer

mean-variance normalization layer

10 years agoMerge pull request #863 from jeffdonahue/lint-check-caffe-fns
Jeff Donahue [Tue, 12 Aug 2014 19:54:29 +0000 (12:54 -0700)]
Merge pull request #863 from jeffdonahue/lint-check-caffe-fns

add lint check for functions with Caffe alternatives (memcpy, memset)

10 years agoFix caffe/alt_fn lint errors.
Jeff Donahue [Wed, 6 Aug 2014 00:33:19 +0000 (17:33 -0700)]
Fix caffe/alt_fn lint errors.

10 years agoCreate caffe_{,gpu_}memset functions to replace {m,cudaM}emset's.
Jeff Donahue [Tue, 5 Aug 2014 23:59:14 +0000 (16:59 -0700)]
Create caffe_{,gpu_}memset functions to replace {m,cudaM}emset's.

10 years agoAdd caffe/alt_fn rule to lint checks to check for functions (like memset
Jeff Donahue [Tue, 5 Aug 2014 23:01:11 +0000 (16:01 -0700)]
Add caffe/alt_fn rule to lint checks to check for functions (like memset
& memcpy) with caffe_* alternatives that should be used instead.

10 years agolint targets should depend on the lint script itself
Jeff Donahue [Tue, 5 Aug 2014 23:19:31 +0000 (16:19 -0700)]
lint targets should depend on the lint script itself

10 years ago[examples] fix links in feature extraction
Evan Shelhamer [Mon, 11 Aug 2014 23:40:02 +0000 (16:40 -0700)]
[examples] fix links in feature extraction

10 years agoMerge pull request #910 from sergeyk/dev
Evan Shelhamer [Mon, 11 Aug 2014 23:11:58 +0000 (16:11 -0700)]
Merge pull request #910 from sergeyk/dev

web demo updates

10 years ago[docs] ‘maximally accurate’ in the web demo explanation. closes #905
Sergey Karayev [Mon, 11 Aug 2014 22:59:48 +0000 (15:59 -0700)]
[docs] ‘maximally accurate’ in the web demo explanation. closes #905

10 years ago[docs] [fix] closes #899
Sergey Karayev [Mon, 11 Aug 2014 22:52:56 +0000 (15:52 -0700)]
[docs] [fix] closes #899

10 years agoMerge pull request #909 from sergeyk/dev
Evan Shelhamer [Mon, 11 Aug 2014 22:17:46 +0000 (15:17 -0700)]
Merge pull request #909 from sergeyk/dev

[docs] sort examples

10 years ago[docs] sorting of examples. if doesn’t work for you, update jekyll.
Sergey Karayev [Mon, 11 Aug 2014 22:09:53 +0000 (15:09 -0700)]
[docs] sorting of examples. if doesn’t work for you, update jekyll.

10 years agolint
qipeng [Mon, 11 Aug 2014 18:09:57 +0000 (11:09 -0700)]
lint

10 years agominor fix for layer factory
qipeng [Mon, 11 Aug 2014 17:48:57 +0000 (10:48 -0700)]
minor fix for layer factory

10 years agoadded cross-channel MVN, Mean-only normalization, added to layer factory, moved to...
qipeng [Mon, 11 Aug 2014 17:44:27 +0000 (10:44 -0700)]
added cross-channel MVN, Mean-only normalization, added to layer factory, moved to common_layers

10 years agoMerge pull request #904 from Yangqing/sweep
Yangqing Jia [Mon, 11 Aug 2014 17:35:04 +0000 (10:35 -0700)]
Merge pull request #904 from Yangqing/sweep

Add a leveldb option function in io.hpp/cpp

10 years agoaddressed Jeff's comment
qipeng [Mon, 4 Aug 2014 18:29:54 +0000 (11:29 -0700)]
addressed Jeff's comment

10 years agolint
qipeng [Mon, 4 Aug 2014 16:27:14 +0000 (09:27 -0700)]
lint

10 years agoreduced blas calls
qipeng [Mon, 4 Aug 2014 16:03:34 +0000 (09:03 -0700)]
reduced blas calls

10 years agomean-variance normalization layer
qipeng [Mon, 4 Aug 2014 01:19:58 +0000 (18:19 -0700)]
mean-variance normalization layer

10 years agoAdd a leveldb option function in io.hpp/cpp
Yangqing Jia [Mon, 11 Aug 2014 16:16:51 +0000 (09:16 -0700)]
Add a leveldb option function in io.hpp/cpp

10 years agofix formatting error in blob.hpp
Jonathan L Long [Mon, 11 Aug 2014 04:23:34 +0000 (21:23 -0700)]
fix formatting error in blob.hpp

10 years agooutput loss for caffenet and alexnet train/val models
Evan Shelhamer [Sun, 10 Aug 2014 05:24:15 +0000 (22:24 -0700)]
output loss for caffenet and alexnet train/val models

10 years agodefault raw_scale in python scripts to ImageNet model value
Evan Shelhamer [Sun, 10 Aug 2014 03:43:40 +0000 (20:43 -0700)]
default raw_scale in python scripts to ImageNet model value

10 years agoMerge pull request #891 from kloudkl/gflags_namespace
Evan Shelhamer [Sat, 9 Aug 2014 21:20:44 +0000 (14:20 -0700)]
Merge pull request #891 from kloudkl/gflags_namespace

Fix the gflags namespace issue

10 years agoMerge pull request #892 from kloudkl/distribute_generated_proto_headers
Jeff Donahue [Sat, 9 Aug 2014 19:43:42 +0000 (12:43 -0700)]
Merge pull request #892 from kloudkl/distribute_generated_proto_headers

Distribute the generated proto header files

10 years agoDistribute the generated proto header files
Kai Li [Sat, 9 Aug 2014 12:41:58 +0000 (20:41 +0800)]
Distribute the generated proto header files

10 years agoFix the gflags namespace issue
Kai Li [Sat, 9 Aug 2014 12:11:38 +0000 (20:11 +0800)]
Fix the gflags namespace issue

10 years agoMerge pull request #888 from ronghanghu/matcaffe-add-check
Sergio Guadarrama [Sat, 9 Aug 2014 09:29:05 +0000 (02:29 -0700)]
Merge pull request #888 from ronghanghu/matcaffe-add-check

add necessary input checks for matcaffe

10 years agoadd necessary input checks for matcaffe
Ronghang Hu [Fri, 8 Aug 2014 22:47:23 +0000 (15:47 -0700)]
add necessary input checks for matcaffe

10 years ago[example] fix example names
Evan Shelhamer [Fri, 8 Aug 2014 19:40:42 +0000 (12:40 -0700)]
[example] fix example names

10 years ago[docs] fix find complaint in example gathering script
Evan Shelhamer [Fri, 8 Aug 2014 19:01:39 +0000 (12:01 -0700)]
[docs] fix find complaint in example gathering script

10 years ago[example] fix broken links in ImageNet recipe
Evan Shelhamer [Fri, 8 Aug 2014 18:18:46 +0000 (11:18 -0700)]
[example] fix broken links in ImageNet recipe

10 years agoBack-merge documentation and fixes
Evan Shelhamer [Thu, 7 Aug 2014 21:17:48 +0000 (14:17 -0700)]
Back-merge documentation and fixes

10 years agoMerge pull request #872 from shelhamer/caffe-tool
Jeff Donahue [Thu, 7 Aug 2014 20:01:14 +0000 (13:01 -0700)]
Merge pull request #872 from shelhamer/caffe-tool

Improve caffe tool

10 years agoconsolidate gpu and device_id args in caffe tool
Evan Shelhamer [Thu, 7 Aug 2014 18:56:45 +0000 (11:56 -0700)]
consolidate gpu and device_id args in caffe tool

10 years agoupdate cli usage in examples
Evan Shelhamer [Thu, 7 Aug 2014 06:56:06 +0000 (23:56 -0700)]
update cli usage in examples

10 years agofix deprecation warnings
Evan Shelhamer [Thu, 7 Aug 2014 06:48:29 +0000 (23:48 -0700)]
fix deprecation warnings

10 years agoconsolidate test into caffe cli
Evan Shelhamer [Thu, 7 Aug 2014 06:38:17 +0000 (23:38 -0700)]
consolidate test into caffe cli

10 years agocomment caffe cli
Evan Shelhamer [Thu, 7 Aug 2014 06:22:56 +0000 (23:22 -0700)]
comment caffe cli

10 years agocheck required caffe cli args
Evan Shelhamer [Thu, 7 Aug 2014 06:22:40 +0000 (23:22 -0700)]
check required caffe cli args

10 years agorename caffe cli args and revise text
Evan Shelhamer [Thu, 7 Aug 2014 06:22:13 +0000 (23:22 -0700)]
rename caffe cli args and revise text

10 years agogive usage message for caffe cli
Evan Shelhamer [Thu, 7 Aug 2014 03:17:00 +0000 (20:17 -0700)]
give usage message for caffe cli

- call format
- commands
- flags

10 years agooutput INFO from caffe cli to stderr by default
Evan Shelhamer [Thu, 7 Aug 2014 03:10:27 +0000 (20:10 -0700)]
output INFO from caffe cli to stderr by default

10 years agoconsolidaet GPU flag for caffe cli
Evan Shelhamer [Thu, 7 Aug 2014 02:45:05 +0000 (19:45 -0700)]
consolidaet GPU flag for caffe cli

10 years agorename tools
Evan Shelhamer [Thu, 7 Aug 2014 02:22:58 +0000 (19:22 -0700)]
rename tools

10 years agoPainless binary mean conversion to matlab matrices.
Alireza Shafaei [Wed, 6 Aug 2014 04:34:45 +0000 (21:34 -0700)]
Painless binary mean conversion to matlab matrices.

10 years agoMerge pull request #868 from shelhamer/license-copyright
Evan Shelhamer [Thu, 7 Aug 2014 01:39:24 +0000 (18:39 -0700)]
Merge pull request #868 from shelhamer/license-copyright

Clarify the license and copyright terms of the project

10 years agolint for copyright
Evan Shelhamer [Wed, 6 Aug 2014 07:36:00 +0000 (00:36 -0700)]
lint for copyright

10 years ago[docs] detail attribution, license, and copyright for development
Evan Shelhamer [Wed, 6 Aug 2014 07:34:05 +0000 (00:34 -0700)]
[docs] detail attribution, license, and copyright for development

10 years agoLICENSE governs the whole project so strip file headers
Evan Shelhamer [Wed, 6 Aug 2014 07:43:17 +0000 (00:43 -0700)]
LICENSE governs the whole project so strip file headers

10 years agoclarify the license and copyright terms of the project
Evan Shelhamer [Wed, 6 Aug 2014 07:16:36 +0000 (00:16 -0700)]
clarify the license and copyright terms of the project

As has been the case, contributions are copyright their respective
contributors and the project is BSD-2 licensed. By contributing to the
project, contributors release their contributions under these copyright
and license terms as declared in LICENSE.

10 years agoMerge pull request #816 from shelhamer/pycaffe-labels-grayscale-attrs-examples
Evan Shelhamer [Wed, 6 Aug 2014 06:24:41 +0000 (23:24 -0700)]
Merge pull request #816 from shelhamer/pycaffe-labels-grayscale-attrs-examples

Improve and polish pycaffe

10 years agodrop np.asarray() in favor of declaration (~1.75x speedup)
Evan Shelhamer [Tue, 5 Aug 2014 17:14:35 +0000 (10:14 -0700)]
drop np.asarray() in favor of declaration (~1.75x speedup)

10 years agofix pycaffe context cropping with or without mean
Evan Shelhamer [Sat, 2 Aug 2014 01:46:17 +0000 (18:46 -0700)]
fix pycaffe context cropping with or without mean

10 years agotake array in pycaffe `Net.set_mean()` instead of file path
Evan Shelhamer [Sat, 2 Aug 2014 00:55:30 +0000 (17:55 -0700)]
take array in pycaffe `Net.set_mean()` instead of file path

10 years agofix pycaffe input processing
Evan Shelhamer [Thu, 31 Jul 2014 23:19:20 +0000 (16:19 -0700)]
fix pycaffe input processing

- load an image as [0,1] single / np.float32 according to Python convention
- fix input scaling during preprocessing:
  - scale input for preprocessing by `raw_scale` e.g. to map an image
    to [0, 255] for the CaffeNet and AlexNet ImageNet models
  - scale feature space by `input_scale` after mean subtraction
  - switch examples to raw scale for ImageNet models
  - fix #525
- preserve type after resizing.
- resize 1, 3, or K channel images with special casing between
  skimage.transform (1 and 3) and scipy.ndimage (K) for speed

10 years ago[example] include prediction in classification, time on GTX 770
Evan Shelhamer [Tue, 29 Jul 2014 03:16:52 +0000 (20:16 -0700)]
[example] include prediction in classification, time on GTX 770

10 years ago[example] fix example outputs
Evan Shelhamer [Mon, 28 Jul 2014 23:08:44 +0000 (16:08 -0700)]
[example] fix example outputs

With the right input processing, the actual image classification output
is sensible.

- filter visualization example's top prediction is "tabby cat"
- net surgery fully-convolutional output map is better

Fix incorrect class names too.

10 years ago[example] add caffe to pythonpath in all examples
Evan Shelhamer [Tue, 29 Jul 2014 03:10:16 +0000 (20:10 -0700)]
[example] add caffe to pythonpath in all examples

10 years agodefine caffe.Net input preprocessing members by boost::python
Evan Shelhamer [Mon, 28 Jul 2014 21:55:13 +0000 (14:55 -0700)]
define caffe.Net input preprocessing members by boost::python

define `Net.{mean, input_scale, channel_swap}` on the boost::python side
so that the members always exist. drop ugly initialization logic.

10 years agoMerge pull request #856 from jeffdonahue/lint-tweaks
Jeff Donahue [Tue, 5 Aug 2014 17:51:07 +0000 (10:51 -0700)]
Merge pull request #856 from jeffdonahue/lint-tweaks

add header alphabetization to lint checks

10 years agoMerge pull request #859 from beam2d/fix-cifar-lrn-region
Jeff Donahue [Tue, 5 Aug 2014 16:22:49 +0000 (09:22 -0700)]
Merge pull request #859 from beam2d/fix-cifar-lrn-region

Fix conflict on setting of LRN layers between train/test net and deploy net

10 years agoFix conflict on setting of LRN layers between train/test net and deploy net
Seiya Tokui [Tue, 5 Aug 2014 09:07:19 +0000 (18:07 +0900)]
Fix conflict on setting of LRN layers between train/test net and deploy net

10 years agoFix header alphabetization lint errors.
Jeff Donahue [Sat, 2 Aug 2014 03:14:42 +0000 (20:14 -0700)]
Fix header alphabetization lint errors.

10 years agoEnable the 'build/include_alpha' rules to make lint check that the
Jeff Donahue [Sat, 2 Aug 2014 02:58:56 +0000 (19:58 -0700)]
Enable the 'build/include_alpha' rules to make lint check that the
includes are correctly alphabetized.

10 years agoAdd "lintclean" target to remove current lint outputs -- forces lint to
Jeff Donahue [Sat, 2 Aug 2014 02:42:41 +0000 (19:42 -0700)]
Add "lintclean" target to remove current lint outputs -- forces lint to
be run again next time "make lint" is run.

10 years agoChanged path in description to point to parse_log.sh
Vik Kamath [Mon, 7 Jul 2014 12:01:56 +0000 (15:01 +0300)]
Changed path in description to point to parse_log.sh

It seems that parse_log.sh has moved from /path/to/caffe/scripts/ to /path/to/caffe/tools/extra. The edit changes the description to reflect this change. Nothing major really.

10 years agoMerge pull request #855 from jeffdonahue/sgd-solver-test
Jeff Donahue [Mon, 4 Aug 2014 23:04:38 +0000 (16:04 -0700)]
Merge pull request #855 from jeffdonahue/sgd-solver-test

SGDSolver tests

10 years agoAdd (momentum) SGD solver tests to check that learning rate, weight
Jeff Donahue [Sun, 3 Aug 2014 23:00:46 +0000 (16:00 -0700)]
Add (momentum) SGD solver tests to check that learning rate, weight
decay, and momentum are implemented properly on a least squares problem.

10 years agoAdd 'snapshot_after_train' to SolverParameter to override the final
Jeff Donahue [Mon, 4 Aug 2014 20:09:00 +0000 (13:09 -0700)]
Add 'snapshot_after_train' to SolverParameter to override the final
snapshot.

10 years agomodified test_concat_layer.cpp
zszhong [Sun, 3 Aug 2014 05:55:06 +0000 (13:55 +0800)]
modified test_concat_layer.cpp

10 years agoMerge pull request #848 from netheril96/random
Jeff Donahue [Mon, 4 Aug 2014 19:57:17 +0000 (12:57 -0700)]
Merge pull request #848 from netheril96/random

Fix and improve multiple places about random number generation and shuffling

10 years agoFix and improve multiple places about random number generation and
netheril96 [Mon, 4 Aug 2014 04:28:21 +0000 (12:28 +0800)]
Fix and improve multiple places about random number generation and
shuffling

10 years agoMerge pull request #834 from qipeng/dev
Yangqing Jia [Sun, 3 Aug 2014 15:53:31 +0000 (08:53 -0700)]
Merge pull request #834 from qipeng/dev

Included cmath in common.hpp to avoid isnan complaints from the xcode co...

10 years agoMerge pull request #840 from jeffdonahue/fix-net-speedtest
Jeff Donahue [Sat, 2 Aug 2014 20:43:39 +0000 (13:43 -0700)]
Merge pull request #840 from jeffdonahue/fix-net-speedtest

Fix speed test tool (and possibly others); add script to speedtest imagenet

10 years agoFix speedtest (and possibly other tools) by setting the net phase to the
Jeff Donahue [Sat, 2 Aug 2014 20:17:15 +0000 (13:17 -0700)]
Fix speedtest (and possibly other tools) by setting the net phase to the
current Caffe::phase() unless explicitly specified in the state.

10 years agoAdd tests for phase filtering according to Caffe singleton phase
Jeff Donahue [Sat, 2 Aug 2014 20:10:46 +0000 (13:10 -0700)]
Add tests for phase filtering according to Caffe singleton phase
(currently failing as FilterNet ignores the singleton phase).

10 years agoAdd script to speedtest imagenet (currently broken as FilterNet ignores
Jeff Donahue [Sat, 2 Aug 2014 20:17:09 +0000 (13:17 -0700)]
Add script to speedtest imagenet (currently broken as FilterNet ignores
Caffe::phase()).

10 years agoMerge pull request #839 from jeffdonahue/test-dropout-ratio
Jeff Donahue [Sat, 2 Aug 2014 03:01:12 +0000 (20:01 -0700)]
Merge pull request #839 from jeffdonahue/test-dropout-ratio

Test that DropoutLayer obeys dropout_ratio

10 years agoTest that DropoutLayer obeys dropout_ratio.
Jeff Donahue [Sat, 2 Aug 2014 02:10:16 +0000 (19:10 -0700)]
Test that DropoutLayer obeys dropout_ratio.