Skip to content

Commit 8ef9213

Browse files
Release 2018.06 RC1
1 parent 194cc0a commit 8ef9213

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

62 files changed

+40548
-0
lines changed
Lines changed: 53 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,53 @@
1+
@set TOOL=evgencnn
2+
@set IMAGES=/local/one_image
3+
@set TOOLDIR=N:\git\cnn\tools\cnn_tools\evgencnn\scripts
4+
set show_classify=1
5+
@set __SAVER=1
6+
@set g_fixed_float=1
7+
@set W=no_weight_file
8+
@set W=dummy_file
9+
@set W=caffe_model\resnet101_faster_rcnn_bn_scale_merged_end2end_iter_70000.caffemodel
10+
@set ROI_TRACE=1
11+
@set RPN_POST_NMS_TOP_N=10
12+
13+
@call :doit 12
14+
@goto exit
15+
@call :doit 8
16+
@call :doit 7
17+
@call :doit 9
18+
@call :doit 10
19+
@call :doit 11
20+
21+
:doit
22+
@set SZ=%1
23+
@set WSZ=%1
24+
python %TOOLDIR%/evgencnn ^
25+
--caffe caffe_model/test.pt ^
26+
--weights %W% ^
27+
--images %IMAGES% ^
28+
--name test ^
29+
--wof bin --vof bin --name test ^
30+
--fpsize %SZ% --fixed_dir fp.%SZ% --fpweight_size %WSZ% ^
31+
--float_dir flt ^
32+
--outdir . ^
33+
--allow_unsupported_hardware ^
34+
--ignore_unsupported_layers ^
35+
--float_type float ^
36+
--signed_blobs ^
37+
-g host_float ^
38+
-g host_fixed ^
39+
--ibs 8 ^
40+
--subgraph G1 proposal cls_score bbox_pred ^
41+
--batch_iterate G1 iobj %RPN_POST_NMS_TOP_N% ^
42+
--pixel_mean 102.9801,115.9465,122.7717 ^
43+
%9
44+
@goto :EOF
45+
46+
@goto end
47+
-g host_fixed ^
48+
-g host_float ^
49+
--no_verify_blobs ^
50+
:end
51+
52+
:exit
53+
@find . -rmdir -s
Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,3 @@
1+
version https://git-lfs.github.com/spec/v1
2+
oid sha256:e1c008d82b9ba458b685a00074c4a2324d1dd0e4bf98eb43d7f17518d4ef44ff
3+
size 548317115
Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,9 @@
1+
The prototxt was taken from
2+
3+
models/pascal_voc/ResNet101_BN_SCALE_Merged/faster_rcnn_end2end/test.prototxt
4+
and modified to allow reasonable generation of dummy weights.
5+
6+
The repository from which this was taken is:
7+
8+
https://github.com/Eniac-Xie/faster-rcnn-resnet
9+
~

0 commit comments

Comments
 (0)