Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
s920243400
PaddleDetection
提交
f7890b1b
P
PaddleDetection
项目概览
s920243400
/
PaddleDetection
与 Fork 源项目一致
Fork自
PaddlePaddle / PaddleDetection
通知
2
Star
0
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
PaddleDetection
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
未验证
提交
f7890b1b
编写于
8月 02, 2022
作者:
S
shangliang Xu
提交者:
GitHub
8月 02, 2022
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
[TIPC] fix pose benchmark shell and txt (#6564)
上级
0d481790
变更
23
隐藏空白更改
内联
并排
Showing
23 changed file
with
78 addition
and
65 deletion
+78
-65
ppdet/data/source/dataset.py
ppdet/data/source/dataset.py
+2
-4
test_tipc/benchmark_train.sh
test_tipc/benchmark_train.sh
+1
-0
test_tipc/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco_train_infer_python.txt
...r_rcnn/faster_rcnn_r50_fpn_1x_coco_train_infer_python.txt
+1
-1
test_tipc/configs/fcos/fcos_r50_fpn_1x_coco_train_infer_python.txt
.../configs/fcos/fcos_r50_fpn_1x_coco_train_infer_python.txt
+1
-1
test_tipc/configs/gfl/gfl_r50_fpn_1x_coco_train_infer_python.txt
...pc/configs/gfl/gfl_r50_fpn_1x_coco_train_infer_python.txt
+1
-1
test_tipc/configs/keypoint/higherhrnet_hrnet_w32_512_train_infer_python.txt
...keypoint/higherhrnet_hrnet_w32_512_train_infer_python.txt
+1
-1
test_tipc/configs/keypoint/hrnet_w32_256x192_train_infer_python.txt
...configs/keypoint/hrnet_w32_256x192_train_infer_python.txt
+1
-1
test_tipc/configs/keypoint/tinypose_128x96_train_infer_python.txt
...c/configs/keypoint/tinypose_128x96_train_infer_python.txt
+4
-4
test_tipc/configs/mask_rcnn/mask_rcnn_r50_1x_coco_train_infer_python.txt
...gs/mask_rcnn/mask_rcnn_r50_1x_coco_train_infer_python.txt
+7
-0
test_tipc/configs/mask_rcnn/mask_rcnn_r50_fpn_1x_coco_train_infer_python.txt
...ask_rcnn/mask_rcnn_r50_fpn_1x_coco_train_infer_python.txt
+1
-1
test_tipc/configs/mot/fairmot_dla34_30e_1088x608_train_infer_python.txt
...igs/mot/fairmot_dla34_30e_1088x608_train_infer_python.txt
+2
-2
test_tipc/configs/mot/jde_darknet53_30e_1088x608_train_infer_python.txt
...igs/mot/jde_darknet53_30e_1088x608_train_infer_python.txt
+1
-1
test_tipc/configs/picodet/picodet_lcnet_1_5x_416_coco_train_infer_python.txt
...icodet/picodet_lcnet_1_5x_416_coco_train_infer_python.txt
+4
-4
test_tipc/configs/picodet/picodet_s_320_coco_lcnet_train_infer_python.txt
...s/picodet/picodet_s_320_coco_lcnet_train_infer_python.txt
+6
-6
test_tipc/configs/picodet/picodet_s_320_coco_train_infer_python.txt
...configs/picodet/picodet_s_320_coco_train_infer_python.txt
+4
-4
test_tipc/configs/ppyolo/ppyolo_mbv3_large_coco_train_infer_python.txt
...figs/ppyolo/ppyolo_mbv3_large_coco_train_infer_python.txt
+4
-4
test_tipc/configs/ppyolo/ppyolo_r50vd_dcn_1x_coco_train_infer_python.txt
...gs/ppyolo/ppyolo_r50vd_dcn_1x_coco_train_infer_python.txt
+4
-4
test_tipc/configs/ppyolo/ppyolo_tiny_650e_coco_train_infer_python.txt
...nfigs/ppyolo/ppyolo_tiny_650e_coco_train_infer_python.txt
+4
-4
test_tipc/configs/ppyolo/ppyolov2_r50vd_dcn_365e_coco_train_infer_python.txt
...pyolo/ppyolov2_r50vd_dcn_365e_coco_train_infer_python.txt
+4
-4
test_tipc/configs/ppyoloe/ppyoloe_crn_s_300e_coco_train_infer_python.txt
...gs/ppyoloe/ppyoloe_crn_s_300e_coco_train_infer_python.txt
+4
-4
test_tipc/configs/solov2/solov2_r50_fpn_1x_coco_train_infer_python.txt
...figs/solov2/solov2_r50_fpn_1x_coco_train_infer_python.txt
+1
-1
test_tipc/configs/yolov3/yolov3_darknet53_270e_coco_train_infer_python.txt
.../yolov3/yolov3_darknet53_270e_coco_train_infer_python.txt
+1
-1
test_tipc/prepare.sh
test_tipc/prepare.sh
+19
-12
未找到文件。
ppdet/data/source/dataset.py
浏览文件 @
f7890b1b
...
...
@@ -68,20 +68,18 @@ class DetDataset(Dataset):
return
self
def
__getitem__
(
self
,
idx
):
n
=
len
(
self
.
roidbs
)
if
self
.
repeat
>
1
:
idx
%=
self
.
repeat
idx
%=
n
# data batch
roidb
=
copy
.
deepcopy
(
self
.
roidbs
[
idx
])
if
self
.
mixup_epoch
==
0
or
self
.
_epoch
<
self
.
mixup_epoch
:
n
=
len
(
self
.
roidbs
)
idx
=
np
.
random
.
randint
(
n
)
roidb
=
[
roidb
,
copy
.
deepcopy
(
self
.
roidbs
[
idx
])]
elif
self
.
cutmix_epoch
==
0
or
self
.
_epoch
<
self
.
cutmix_epoch
:
n
=
len
(
self
.
roidbs
)
idx
=
np
.
random
.
randint
(
n
)
roidb
=
[
roidb
,
copy
.
deepcopy
(
self
.
roidbs
[
idx
])]
elif
self
.
mosaic_epoch
==
0
or
self
.
_epoch
<
self
.
mosaic_epoch
:
n
=
len
(
self
.
roidbs
)
roidb
=
[
roidb
,
]
+
[
copy
.
deepcopy
(
self
.
roidbs
[
np
.
random
.
randint
(
n
)])
for
_
in
range
(
4
)
...
...
test_tipc/benchmark_train.sh
浏览文件 @
f7890b1b
...
...
@@ -166,6 +166,7 @@ else
fi
if
[[
${
model_name
}
=
~
"higherhrnet"
]]
||
[[
${
model_name
}
=
~
"hrnet"
]]
||
[[
${
model_name
}
=
~
"tinypose"
]]
;
then
echo
"
${
model_name
}
run on full coco dataset"
epoch
=
$(
set_dynamic_epoch
$device_num
$epoch
)
else
epoch
=
1
...
...
test_tipc/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -53,7 +53,7 @@ inference:./deploy/python/infer.py
batch_size:2|8
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
3
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
===========================infer_benchmark_params===========================
...
...
test_tipc/configs/fcos/fcos_r50_fpn_1x_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -53,7 +53,7 @@ inference:./deploy/python/infer.py
batch_size:2|8
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
3
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
===========================infer_benchmark_params===========================
...
...
test_tipc/configs/gfl/gfl_r50_fpn_1x_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -53,6 +53,6 @@ inference:./deploy/python/infer.py
batch_size:2|8
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
3
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
test_tipc/configs/keypoint/higherhrnet_hrnet_w32_512_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -52,7 +52,7 @@ inference:./deploy/python/keypoint_infer.py
===========================train_benchmark_params==========================
batch_size:20|24
fp_items:fp32|fp16
epoch:
20
epoch:
1
repeat:1
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
...
...
test_tipc/configs/keypoint/hrnet_w32_256x192_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -52,7 +52,7 @@ null:null
===========================train_benchmark_params==========================
batch_size:64|160
fp_items:fp32|fp16
epoch:
40
epoch:
1
repeat:1
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
...
...
test_tipc/configs/keypoint/tinypose_128x96_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/keypoint_infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
random_infer_input:[{float32,[3,128,96]}]
===========================train_benchmark_params==========================
batch_size:512
fp_items:fp32|fp16
epoch:
25
epoch:
1
repeat:1
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
random_infer_input:[{float32,[3,128,96]}]
\ No newline at end of file
test_tipc/configs/mask_rcnn/mask_rcnn_r50_1x_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,5 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
--trt_max_shape:1600
===========================train_benchmark_params==========================
batch_size:2|4
fp_items:fp32|fp16
epoch:1
repeat:2
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x800x1344.npy
\ No newline at end of file
test_tipc/configs/mask_rcnn/mask_rcnn_r50_fpn_1x_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -53,7 +53,7 @@ inference:./deploy/python/infer.py
batch_size:2|4
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
2
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
===========================infer_benchmark_params===========================
...
...
test_tipc/configs/mot/fairmot_dla34_30e_1088x608_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -50,10 +50,10 @@ inference:./deploy/pptracking/python/mot_jde_infer.py
--run_benchmark:False
--trt_max_shape:1600
===========================train_benchmark_params==========================
batch_size:6|2
2
batch_size:6|2
0
fp_items:fp32|fp16
epoch:1
repeat:
2
repeat:
1
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
===========================infer_benchmark_params===========================
...
...
test_tipc/configs/mot/jde_darknet53_30e_1088x608_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -50,7 +50,7 @@ inference:./deploy/pptracking/python/mot_jde_infer.py
--run_benchmark:False
--trt_max_shape:1600
===========================train_benchmark_params==========================
batch_size:4|1
4
batch_size:4|1
2
fp_items:fp32|fp16
epoch:1
repeat:1
...
...
test_tipc/configs/picodet/picodet_lcnet_1_5x_416_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x416x416_2.npy
===========================train_benchmark_params==========================
batch_size:80
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
30
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x416x416_2.npy
\ No newline at end of file
test_tipc/configs/picodet/picodet_s_320_coco_lcnet_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -6,7 +6,7 @@ use_gpu:True
auto_cast:null
epoch:lite_train_lite_infer=1|lite_train_whole_infer=1|whole_train_whole_infer=300
save_dir:null
TrainReader.batch_size:lite_train_lite_infer=2|lite_train_whole_infer=2|whole_train_whole_infer=
128
TrainReader.batch_size:lite_train_lite_infer=2|lite_train_whole_infer=2|whole_train_whole_infer=
64
pretrain_weights:https://paddledet.bj.bcebos.com/models/picodet_s_320_coco_lcnet.pdparams
trained_model_name:model_final.pdparams
train_infer_img_dir:./dataset/coco/test2017/
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320_2.npy
===========================train_benchmark_params==========================
batch_size:
128
batch_size:
64
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
25
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320_2.npy
\ No newline at end of file
test_tipc/configs/picodet/picodet_s_320_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320_2.npy
===========================train_benchmark_params==========================
batch_size:128
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
50
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320_2.npy
\ No newline at end of file
test_tipc/configs/ppyolo/ppyolo_mbv3_large_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320.npy
===========================train_benchmark_params==========================
batch_size:24
fp_items:fp32|fp16
epoch:1
repeat:1
repeat:1
0
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320.npy
\ No newline at end of file
test_tipc/configs/ppyolo/ppyolo_r50vd_dcn_1x_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x608x608.npy
===========================train_benchmark_params==========================
batch_size:24
fp_items:fp32|fp16
epoch:1
repeat:1
repeat:1
0
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x608x608.npy
\ No newline at end of file
test_tipc/configs/ppyolo/ppyolo_tiny_650e_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320.npy
===========================train_benchmark_params==========================
batch_size:32
fp_items:fp32|fp16
epoch:1
repeat:1
repeat:1
2
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x320x320.npy
\ No newline at end of file
test_tipc/configs/ppyolo/ppyolov2_r50vd_dcn_365e_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
null:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x640x640.npy
===========================train_benchmark_params==========================
batch_size:12
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
5
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x640x640.npy
\ No newline at end of file
test_tipc/configs/ppyoloe/ppyoloe_crn_s_300e_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -49,12 +49,12 @@ inference:./deploy/python/infer.py
--save_log_path:null
--run_benchmark:False
--trt_max_shape:1600
===========================infer_benchmark_params===========================
numpy_infer_input:3x640x640.npy
===========================train_benchmark_params==========================
batch_size:32
fp_items:fp32|fp16
epoch:1
repeat:1
repeat:1
2
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
\ No newline at end of file
flags:null
===========================infer_benchmark_params===========================
numpy_infer_input:3x640x640.npy
\ No newline at end of file
test_tipc/configs/solov2/solov2_r50_fpn_1x_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -53,7 +53,7 @@ inference:./deploy/python/infer.py
batch_size:2|4
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
2
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
===========================infer_benchmark_params===========================
...
...
test_tipc/configs/yolov3/yolov3_darknet53_270e_coco_train_infer_python.txt
浏览文件 @
f7890b1b
...
...
@@ -53,7 +53,7 @@ null:null
batch_size:8
fp_items:fp32|fp16
epoch:1
repeat:
1
repeat:
3
--profiler_options:batch_range=[10,20];state=GPU;tracer_option=Default;profile_path=model.profile
flags:null
===========================infer_benchmark_params===========================
...
...
test_tipc/prepare.sh
浏览文件 @
f7890b1b
...
...
@@ -86,18 +86,25 @@ elif [ ${MODE} = "benchmark_train" ];then
pip
install
-U
pip
pip
install
Cython
pip
install
-r
requirements.txt
# prepare lite benchmark coco data
wget
-nc
-P
./dataset/coco/ https://paddledet.bj.bcebos.com/data/coco_benchmark.tar
--no-check-certificate
cd
./dataset/coco/
&&
tar
-xvf
coco_benchmark.tar
mv
-u
coco_benchmark/
*
./
ls
./
cd
../../
# prepare lite benchmark mot data
wget
-nc
-P
./dataset/mot/ https://paddledet.bj.bcebos.com/data/mot_benchmark.tar
--no-check-certificate
cd
./dataset/mot/
&&
tar
-xvf
mot_benchmark.tar
mv
-u
mot_benchmark/
*
./
ls
./
cd
../../
if
[[
${
model_name
}
=
~
"higherhrnet"
]]
||
[[
${
model_name
}
=
~
"hrnet"
]]
||
[[
${
model_name
}
=
~
"tinypose"
]]
;
then
wget
-nc
-P
./dataset/ https://bj.bcebos.com/v1/paddledet/data/coco.tar
--no-check-certificate
cd
./dataset/
&&
tar
-xf
coco.tar
ls
./coco/
cd
../
else
# prepare lite benchmark coco data
wget
-nc
-P
./dataset/coco/ https://paddledet.bj.bcebos.com/data/coco_benchmark.tar
--no-check-certificate
cd
./dataset/coco/
&&
tar
-xf
coco_benchmark.tar
mv
-u
coco_benchmark/
*
./
ls
./
cd
../../
# prepare lite benchmark mot data
wget
-nc
-P
./dataset/mot/ https://paddledet.bj.bcebos.com/data/mot_benchmark.tar
--no-check-certificate
cd
./dataset/mot/
&&
tar
-xf
mot_benchmark.tar
mv
-u
mot_benchmark/
*
./
ls
./
cd
../../
fi
elif
[
${
MODE
}
=
"paddle2onnx_infer"
]
;
then
# install paddle2onnx
${
python
}
-m
pip
install
paddle2onnx
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录