Detection - #13

Open

NourElmadany wants to merge 1092 commits into

MMLab-CU:masterfrom

yjxiong:untrimmednet

This pull request is big! We're only showing the most recent 250 commits

Commits on Jul 6, 2015

minor fixes
yjxiong
committed
0b389ee

Commits on Jul 7, 2015

List protobuf-compiler dependency in the correct place (it is in the package managers for both 14.04 and 12.04)
semitrivial
committed
58b114f
imported video data layer
lmwang
committed
d7407af

Commits on Jul 8, 2015

Fix CMake typos
lukeyeager
committed
a08bfb5
add fixed corner crop support.
yjxiong
committed
ead0392
add fixed corner crop support.
yjxiong
committed
f539b25
Merge remote-tracking branch 'origin/data_parallel' into action_recog
lmwang
committed
b583a4c
Update README.md
yjxiong
committed
582499e
Merge pull request #2716 from lukeyeager/cmake-typos
shelhamer
committed
b41dd87
Making the net_spec python3 compatible
philkr
committed
7093b0b
Merge pull request #2714 from philkr/python3_netspec
longjon
committed
96179ca

Commits on Jul 9, 2015

Update README.md
yjxiong
committed
a97b976
Update README.md
yjxiong
committed
e8a5c4c

Commits on Jul 10, 2015

Update README.md
yjxiong
committed
0e2d7a8
Update README.md
yjxiong
committed
1ef277c
Update README.md
yjxiong
committed
4f008e0
Merge pull request #2712 from semitrivial/master
jeffdonahue
committed
ff4e286
experimental multi-scale cropping support.
yjxiong
committed
3eb1332
Merge branch 'action_recog' of https://github.com/yjxiong/caffe into action_recog
lmwang
committed
b993fda
Merge remote-tracking branch 'origin/data_parallel' into action_recog
lmwang
committed
c5a0e95
multi-scale bug fix
yjxiong
committed
b2742ca
Merge remote-tracking branch 'upstream/master'
yjxiong
committed
0c2b994
Merge branch 'master' into data_parallel
yjxiong
committed
99b4f51
multi-scale bug fix
yjxiong
committed
4ef1979

Commits on Jul 11, 2015

Merge remote-tracking branch 'origin/data_parallel' into action_recog
lmwang
committed
00013c7
multi-scale bug fix
yjxiong
committed
e481636

Commits on Jul 12, 2015

add corner crop and multiscale
lmwang
committed
99c67b4
Merge remote-tracking branch 'origin/data_parallel' into action_recog
lmwang
committed
6950c42
Update README.md
yjxiong
committed
8d54662
remove redundant cropping settings.
yjxiong
committed
d6120bc
resolve conlict of multi-scale and mean file
yjxiong
committed
c7815f9

Commits on Jul 14, 2015

Update README.md
yjxiong
committed
8fbff97
Update README.md
yjxiong
committed
305ce56

Commits on Jul 28, 2015

Merge remote-tracking branch 'origin/data_parallel' into action_recog
wanglimin
committed
6e4ced0

Commits on Jul 29, 2015

add is_flow in data_tranformer layer
lmwang
committed
d7aea0b
modified gather_layer
lmwang
committed
9a9e249
solve the confict
lmwang
committed
0716374

Commits on Jul 30, 2015

add configurable multiscale
yjxiong
committed
ec7b130
Merge remote-tracking branch 'origin/data_parallel' into action_recog
lmwang9527
committed
470b58f

Commits on Jul 31, 2015

Merge branch 'action_recog' of https://github.com/yjxiong/caffe into action_recog
lmwang9527
committed
3e9d3df
resolve merge issues
lmwang9527
committed
ba03b68

Commits on Aug 2, 2015

add example of training very deep two-stream cnns
lmwang
committed
c02f6a1

Commits on Aug 3, 2015

releasing UCF101 model and configurations.
yjxiong
committed
7377a6d
releasing UCF101 model and configurations.
yjxiong
committed
0f59fd3
releasing UCF101 temporal model and configurations.
yjxiong
committed
1607d0d
finish releasing UCF101 models.
yjxiong
committed
13a392e
added proejct site for ucf101 report.
yjxiong
committed
97a32d0
fix typo
yjxiong
committed
f77895b
Update README.md
yjxiong
committed
1ce0fb8
Update train_action_recognition_flow.sh
yjxiong
committed
0e7fae6
Update train_action_recognition_rgb.sh
yjxiong
committed
175c0d3

Commits on Aug 6, 2015

Update vgg_16_rgb_train_val_fast.prototxt
Limin Wang
committed
af00258
Update vgg_16_flow_train_val_fast.prototxt
Limin Wang
committed
8a16357

Commits on Aug 7, 2015

Update readme.md
yjxiong
committed
c55174e

Commits on Aug 11, 2015

add matlab test code for action recognition
lmwang
committed
e41bca5
fix demo.m
lmwang
committed
bea2d3d
update rbg_mean.mat
lmwang
committed
62eb43f
caffe's trial GPU initialization conflict with exclusive mode.
yjxiong
committed
4e4f862
Merge branch 'data_parallel' into action_recog
yjxiong
committed
bef1459
fix rgb_mean.mat
lmwang
committed
fdc32ff
Merge remote-tracking branch 'origin/action_recog' into action_recog
lmwang
committed
3c4ea49

Commits on Aug 14, 2015

solve testing issues
yjxiong
committed
98494fd
Merge branch 'data_parallel' into action_recog
yjxiong
committed
84e80ab
Update README.md
yjxiong
committed
c6f9c47

Commits on Aug 19, 2015

experimental batch normalization
lxx1991
committed
7e974de

Commits on Aug 26, 2015

initial implementation finished.
yjxiong
committed
33af902
background comm implemented. Incorrect perforamance.
yjxiong
committed
fbbd635
cp
yjxiong
committed
6f58275

Commits on Aug 31, 2015

results corrected. locking conflict solved.
yjxiong
committed
8cd6459

Commits on Sep 1, 2015

Update vgg_16_rgb_train_val_fast.prototxt
Limin Wang
committed
aa35a08
possible memleak. iter_size issue
yjxiong
committed
3f30c78
Update vgg_16_flow_train_val_fast.prototxt
Limin Wang
committed
b01756e
Update vgg_16_rgb_train_val_fast.prototxt
Limin Wang
committed
dfd2333
memleak resolved. wrapped code in USE_MPI macro
yjxiong
committed
18608b6
remove redundant code
yjxiong
committed
a4ef029
no need for boost 1.55 for now
yjxiong
committed
99ef931

Commits on Sep 3, 2015

mpi_thread_serialized works properly with current ver.
yjxiong
committed
ac821ab

Commits on Sep 7, 2015

Merge branch 'bg_comm' into data_parallel
yjxiong
committed
218f1d7
Squashed commit of the following:
yjxiong
committed
91ee558
Merge remote-tracking branch 'origin/action_recog' into action_recog
yjxiong
committed
ac117ad
Update README.md
yjxiong
committed
4fafa60
Update README.md
yjxiong
committed
5547704

Commits on Sep 8, 2015

Update README.md
yjxiong
committed
411f38d

Commits on Sep 9, 2015

Update data_transformer.cpp
Limin Wang
committed
bda5241
Update data_transformer.cpp
Limin Wang
committed
4598606
Update README.md
yjxiong
committed
b08c711

Commits on Sep 11, 2015

Update README.md
yjxiong
committed
e111b0a

Commits on Sep 14, 2015

fix Gather layer bug when USE_MPI is disabled.
yjxiong
committed
f5ea4ef
Merge branch 'data_parallel' into action_recog
yjxiong
committed
990a729
Merge remote-tracking branch 'origin/action_recog' into action_recog
yjxiong
committed
032ef41

Commits on Sep 17, 2015

add "exp10" lr_policy
yjxiong
committed
0a03a24
within-shard shuffle implemented.
yjxiong
committed
72a886a

Commits on Sep 18, 2015

minor fixes
yjxiong
committed
77169c4
minor fixes
yjxiong
committed
69944ad
lr_policy typo
yjxiong
committed
354ca5e
disable openmpi cuda api: occasional data corruption.
yjxiong
committed
31d2f55

Commits on Sep 23, 2015

initial upgrade to cudnnv3, fwd only
yjxiong
committed
e3fefd2

Commits on Sep 24, 2015

supported cudnn_v3
yjxiong
committed
a61e075
Merge branch 'cudnn3' into action_recog
yjxiong
committed
ff4e486
Update README.md
yjxiong
committed
6206e6a

Commits on Sep 25, 2015

reduce cuDNN mem consumption in Testing.
yjxiong
committed
3b53d5c
Merge remote-tracking branch 'origin/action_recog' into action_recog
yjxiong
committed
d121be0

Commits on Sep 30, 2015

Update README.md
yjxiong
committed
f0f1dc3
Update README.md
yjxiong
committed
b4c7435

Commits on Oct 5, 2015

refine crop/scale code
yjxiong
committed
2e6c2fa

Commits on Oct 7, 2015

add scatter layer and related logics
yjxiong
committed
88cd584

Commits on Oct 8, 2015

add more cropping options
yjxiong
committed
fe2c9cd

Commits on Oct 9, 2015

try to fix cudnn buffer allocation
yjxiong
committed
b08cfe8

Commits on Oct 10, 2015

Merge branch 'action_recog' into data_parallel
yjxiong
committed
96a7236
refactored cudnn wrapper.
yjxiong
committed
c38a5bd

Commits on Oct 13, 2015

Update README.md
yjxiong
committed
10f17bf

Commits on Oct 16, 2015

add normalize, smoothL1Loss layers
yjxiong
committed
4cdc768
Merge remote-tracking branch 'origin/action_recog' into action_recog
yjxiong
committed
e5f5a37

Commits on Oct 19, 2015

add frozen state to bn for finetuning on small dataset.
yjxiong
committed
99449d8

Commits on Oct 20, 2015

expose phase to python layer
yjxiong
committed
2d3945c

Commits on Oct 22, 2015

force releasing opencv mat
yjxiong
committed
a1585e1
minor change
yjxiong
committed
c30c998

Commits on Nov 11, 2015

add python prefetch thread registry
yjxiong
committed
09ceecf
finish prefetch cpp side
yjxiong
committed
868b3c6

Commits on Nov 13, 2015

Update README.md
yjxiong
committed
a444264
Update README.md
yjxiong
committed
ec7ee95

Commits on Nov 14, 2015

python layer prefetch tested
yjxiong
committed
214bfee

Commits on Nov 19, 2015

Merge branch 'action_recog' of https://github.com/yjxiong/caffe into action_recog
yjxiong
committed
b864508
remove redundant logging
yjxiong
committed
03ed316
remove faulty include
yjxiong
committed
c402f23

Commits on Nov 23, 2015

Update README.md
yjxiong
committed
fb43a7b

Commits on Dec 1, 2015

fix testing bug when crop_size not specified
yjxiong
committed
db9fabe

Commits on Dec 15, 2015

Fix CuDNNConvolutionLayer for cuDNN v4

flx42
authored and
yjxiong
committed
b739b97
Merge branch 'action_recog' of https://github.com/yjxiong/caffe into action_recog
yjxiong
committed
fb0c5da

Commits on Dec 16, 2015

Add cudnn BN
Cysu
committed
0eee271
Python script to convert original BN models to new ones
Cysu
committed
0ade90d
Rename the python script and add a log message
Cysu
committed
83d0315
Merge pull request #16 from Cysu/cudnn-bn
yjxiong
committed
15c86f7

Commits on Dec 17, 2015

Update README.md
yjxiong
committed
a309135
Create CHANGELOG.md
yjxiong
committed
f536113
Update README.md
yjxiong
committed
b5d15a1

Commits on Dec 20, 2015

Find the best cudnn conv algorithm during runtime
Cysu
committed
9d3f308

Commits on Dec 21, 2015

Choose the best cudnn conv algo that satisfies the workspace memory limit
Cysu
committed
31a3bca
Fix a bug
Cysu
committed
c1339de

Commits on Dec 22, 2015

Merge pull request #18 from Cysu/cudnn-find-algo
yjxiong
committed
ec26910
add gloable runtime optimizaiton for cudnn conv
yjxiong
committed
c263c06

Commits on Dec 23, 2015

minor fix for stable optimiz
yjxiong
committed
1e2e358
change mem_tick design
yjxiong
committed
8a61e5c
fix zero richness
yjxiong
committed
d017226
fix zero richness
yjxiong
committed
e34037b
change default richness setting
yjxiong
committed
be6306e
Merge pull request #19 from yjxiong/runtime_optim_cudnn
yjxiong
committed
69f1b6d
Minor change
yjxiong
committed
b927156
Update README.md
yjxiong
committed
e4b9c40
Update CHANGELOG.md
yjxiong
committed
5fa731c
Remove earlier updates
yjxiong
committed
08e8033
Update README.md
yjxiong
committed
10ee85e
Update README.md
yjxiong
committed
5fdfd26
Update CHANGELOG.md
yjxiong
committed
d0677f4
correction of cudnn_conv_layer.cpp
wanglimin
committed
06dc7be
enable travis ci
Yuanjun Xiong
committed
5b9ba73
enable travis ci
Yuanjun Xiong
committed
c5c5d50
fix travis build
yjxiong
committed
fa59f0a
fix travis
yjxiong
committed
991bdcd
fix travis compile failure
yjxiong
committed
876042e
disable lint on travis
yjxiong
committed
7aafb2a
cannot run make runtime on travis with CUDA
yjxiong
committed
31536b8

Commits on Dec 24, 2015

Merge pull request #20 from yjxiong/travis
yjxiong
committed
cc38aa2
Update README.md
yjxiong
committed
efa1289

Commits on Dec 30, 2015

fix restart crash by removing global reference.
yjxiong
committed
e061fd2
Merge pull request #22 from yjxiong/fix/cudnn_lib_crash
yjxiong
committed
973c5d6

Commits on Jan 4, 2016

Commits on Jan 5, 2016

Freeze the BN layer when cannot be absorbed
Cysu
committed
f4ebd91
Merge pull request #26 from Cysu/cudnn-bn
yjxiong
committed
a418398

Commits on Jan 7, 2016

Python wrappers for temporal/spatial prediction
Chuck Cho
committed
f25b523

Commits on Jan 8, 2016

added keyword arguments
Chuck Cho
committed
f34a16c

Commits on Jan 14, 2016

Merge pull request #28 from chuckcho/action_recog_python_wrapper
yjxiong
committed
67904d0
Fix the checking of shared param
Cysu
committed
241f7fd
Merge pull request #32 from Cysu/fix-shared-param
yjxiong
committed
ba053c9

Commits on Jan 16, 2016

Update vgg_16_rgb_train_val_fast.prototxt
wanglimin
committed
e1de4fd
Update vgg_16_flow_train_val_fast.prototxt
wanglimin
committed
aec55f8

Commits on Jan 18, 2016

Update README.md
yjxiong
committed
2fe7caa

Commits on Feb 2, 2016

fallback to caffe's BN engine for accuracy issue.
yjxiong
committed
89bc738

Commits on Feb 17, 2016

Update data_transformer.cpp
wanglimin
committed
a25a24b
Update data_transformer.cpp
wanglimin
committed
2767017

Commits on Feb 19, 2016

Fix bugs in CUDNN BN layer
Cysu
committed
bbedb27
Merge pull request #47 from Cysu/fix-cudnn-bn
Cysu
committed
b4fc873

Commits on Feb 24, 2016

Sync loss when test
Cysu
committed
203f24d

Commits on Feb 25, 2016

Merge pull request #48 from Cysu/fix-test-sync-loss
yjxiong
committed
6ed67a1

Commits on Feb 29, 2016

Update VideoSpatialPrediction.py
yjxiong
committed
4f99d7e
fix broken tests with cuDNN
yjxiong
committed
36bc68a
add support for ROIPooling layers
yjxiong
committed
d8568ae
Merge pull request #50 from yjxiong/fix/cudnn_test
yjxiong
committed
eacddc5
guard roi pooling tests with GPU only macro
yjxiong
committed
3d04a73
Merge branch 'action_recog' of github.com:yjxiong/caffe into roi_pooling_layers
yjxiong
committed
9d1a274
Merge pull request #51 from yjxiong/roi_pooling_layers
yjxiong
committed
3ce4495
Update README.md
yjxiong
committed
5a23d79

Commits on Mar 5, 2016

add scale and bias layers
yjxiong
committed
a5cc263
add scale and bias layers
yjxiong
committed
1ea015c

Commits on Mar 7, 2016

Update video_data_layer.cpp
wanglimin
committed
02c496e

Commits on Mar 8, 2016

add batch red layer
yjxiong
committed
c2622fc
bread tested
Yuanjun Xiong
committed
e629d3e

Commits on Mar 9, 2016

add gpu implementation
yjxiong
committed
51adf35

Commits on Mar 10, 2016

set zero
Yuanjun Xiong
committed
c9ae4f1

Commits on Mar 21, 2016

Merge branch 'action_recog' of https://github.com/yjxiong/caffe into bread
yjxiong
committed
e7ed649

Commits on Mar 30, 2016

add custom file patterns for video data.
yjxiong
committed
20f6070
Update travis_install.sh
yjxiong
committed
e0292c8
Update travis_install.sh
yjxiong
committed
473c9f5
fix cpu build
yjxiong
committed
6fa6382

Commits on Apr 14, 2016

new reduction options.
yjxiong
committed
a4c0c8b

Commits on Apr 15, 2016

fix bugs when the video has only frame
yjxiong
committed
6bd56ae

Commits on Apr 25, 2016

init cudnn_v5 support
yjxiong
committed
8a44477

Commits on Apr 26, 2016

Fix cudnn bn implementation
Cysu
committed
a1bd9af
cpu bn changes. cudnn frozen support.
yjxiong
committed
04495ed
remove older bn related changes.
yjxiong
committed
6baf2e8
Fix frozen backward
Cysu
committed
686e1bc
Unit test for BN and cudnn BN layers
Cysu
committed
0074a02

Commits on Apr 27, 2016

update bn_convert script
yjxiong
committed
82737f5
typo
yjxiong
committed
de7eb2e
typo
yjxiong
committed
e2ebce1
add epsilon option
yjxiong
committed
f519bcf
enable google-style data augmentation.
yjxiong
committed
2f14344
Cudnn v5 Support (#71)
yjxiong
committed
89777d9

Commits on Apr 28, 2016

Add BN style option for gen_bn_inference
Cysu
committed
026da79
use our BN engine in frozen model
yjxiong
committed
77d0cf6
Merge branch 'action_recog' of https://github.com/yjxiong/caffe into action_recog
yjxiong
committed
ed2cc56

Commits on May 10, 2016

init mem optimizer
yjxiong
committed
a15058b
fix share diff cases.
yjxiong
committed
201fcd3
share conv buffer and use fastest mode
yjxiong
committed
c12d530

Commits on May 17, 2016

Merge remote-tracking branch 'origin/bread' into new_anet
yjxiong
committed
5fd5910
fix topk bug
yjxiong
committed
fe898dd

Commits on May 18, 2016

fix snapshot and mem opt time
yjxiong
committed
331349f

Commits on May 20, 2016

Update solver.cpp to fix the duplicate snapshot (#78)

Cysu
authored and
yjxiong
committed
b67600b
Merge remote-tracking branch 'origin/action_recog' into anet
yjxiong
committed
eca349b
improve mem
yjxiong
committed
8d9fca8

Commits on Jun 10, 2016

Update README.md
yjxiong
authored
f02f23e

Commits on Jun 15, 2016

Improve the python interface (#80)

Cysu
authored and
yjxiong
committed
b301fa0

Commits on Jul 16, 2016

Update README.md
yjxiong
authored
b5e2b3c

Commits on Jul 20, 2016

Reuse data memory if not propagate down (#86)

Cysu
authored and
yjxiong
committed
823a7c7
readme and comments
yjxiong
committed
be90db6
Merge remote-tracking branch 'origin/org_size_processing' into mem
yjxiong
committed
3721b02
more comments
yjxiong
committed
34665f5

Commits on Jul 29, 2016

Update README.md (#95)

ae86208
authored and
yjxiong
committed
c0d6a03

Commits on Jul 30, 2016

Update README.md
yjxiong
authored
587df4d
add TOC
yjxiong
authored
42b9db8

Commits on Aug 4, 2016

fix cudnn conv legacy bug (#96)
yjxiong
authored
c0d200b
Merge branch 'action_recog' into mem
yjxiong
committed
70cb98d
Merge branch 'action_recog' of https://github.com/yjxiong/caffe into mem
yjxiong
committed
f971cdb

Commits on Aug 11, 2016

standardize memory optimization configurations (#99)
yjxiong
authored
4506029

Commits on Sep 2, 2016

Split MPI comm by specified groups (#101)

Cysu
authored and
yjxiong
committed
f1fb673

Commits on Aug 19, 2017

untrimmednet push
wanglimin
committed
08c6c8c