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
PaddlePaddle
PaddleClas
提交
76c3f1c6
P
PaddleClas
项目概览
PaddlePaddle
/
PaddleClas
接近 2 年 前同步成功
通知
116
Star
4999
Fork
1114
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
19
列表
看板
标记
里程碑
合并请求
6
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
PaddleClas
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
19
Issue
19
列表
看板
标记
里程碑
合并请求
6
合并请求
6
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
76c3f1c6
编写于
8月 19, 2022
作者:
H
HydrogenSulfate
提交者:
GitHub
8月 19, 2022
浏览文件
操作
浏览文件
下载
差异文件
Merge pull request #2213 from HydrogenSulfate/refine_pksampler
add assertion for DistributedRandomIdentitySampler
上级
da70ef9c
a646adf7
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
26 addition
and
24 deletion
+26
-24
ppcls/data/dataloader/DistributedRandomIdentitySampler.py
ppcls/data/dataloader/DistributedRandomIdentitySampler.py
+12
-9
ppcls/data/dataloader/pk_sampler.py
ppcls/data/dataloader/pk_sampler.py
+14
-15
未找到文件。
ppcls/data/dataloader/DistributedRandomIdentitySampler.py
浏览文件 @
76c3f1c6
...
...
@@ -14,24 +14,27 @@
from
__future__
import
absolute_import
from
__future__
import
division
from
collections
import
defaultdict
import
numpy
as
np
import
copy
import
random
from
collections
import
defaultdict
import
numpy
as
np
from
paddle.io
import
DistributedBatchSampler
,
Sampler
class
DistributedRandomIdentitySampler
(
DistributedBatchSampler
):
"""
Randomly sample N identities, then for each identity,
randomly sample K instances, therefore batch size is N*K.
"""Randomly sample N identities, then for each identity,
randomly sample K instances, therefore batch size equals to N * K.
Args:
- data_source (list): list of (img_path, pid, camid).
- num_instances (int): number of instances per identity in a batch.
- batch_size (int): number of examples in a batch.
dataset(Dataset): Dataset which contains list of (img_path, pid, camid))
batch_size (int): batch size
num_instances (int): number of instance(s) within an class
drop_last (bool): whether to discard the data at the end
"""
def
__init__
(
self
,
dataset
,
batch_size
,
num_instances
,
drop_last
,
**
args
):
assert
batch_size
%
num_instances
==
0
,
\
f
"batch_size(
{
batch_size
}
) must be divisible by num_instances(
{
num_instances
}
) when using DistributedRandomIdentitySampler"
self
.
dataset
=
dataset
self
.
batch_size
=
batch_size
self
.
num_instances
=
num_instances
...
...
ppcls/data/dataloader/pk_sampler.py
浏览文件 @
76c3f1c6
...
...
@@ -14,27 +14,27 @@
from
__future__
import
absolute_import
from
__future__
import
division
from
collections
import
defaultdict
import
numpy
as
np
import
random
from
paddle.io
import
DistributedBatchSampler
from
ppcls.utils
import
logger
class
PKSampler
(
DistributedBatchSampler
):
"""
First, randomly sample P identiti
es.
Then for each identity randomly sample K instances
.
Therefore batch size is P*K, and the sampler called PKSampler.
"""
First, randomly sample P identities.
Then for each identity randomly sample K instanc
es.
Therefore batch size equals to P * K, and the sampler called PKSampler
.
Args:
dataset (paddle.io.Dataset): list of (img_path, pid, cam_id).
sample_per_id(int): number of instances per identity in a batch.
batch_size (int): number of examples in a batch.
shuffle(bool): whether to shuffle indices order before generating
batch indices. Default False.
dataset (Dataset): Dataset which contains list of (img_path, pid, camid))
batch_size (int): batch size
sample_per_id (int): number of instance(s) within an class
shuffle (bool, optional): _description_. Defaults to True.
drop_last (bool, optional): whether to discard the data at the end. Defaults to True.
sample_method (str, optional): sample method when generating prob_list. Defaults to "sample_avg_prob".
"""
def
__init__
(
self
,
dataset
,
batch_size
,
...
...
@@ -42,10 +42,9 @@ class PKSampler(DistributedBatchSampler):
shuffle
=
True
,
drop_last
=
True
,
sample_method
=
"sample_avg_prob"
):
super
().
__init__
(
dataset
,
batch_size
,
shuffle
=
shuffle
,
drop_last
=
drop_last
)
super
().
__init__
(
dataset
,
batch_size
,
shuffle
=
shuffle
,
drop_last
=
drop_last
)
assert
batch_size
%
sample_per_id
==
0
,
\
"PKSampler configs error, Sample_per_id must be a divisor of batch_size
."
f
"PKSampler configs error, sample_per_id(
{
sample_per_id
}
) must be a divisor of batch_size(
{
batch_size
}
)
."
assert
hasattr
(
self
.
dataset
,
"labels"
),
"Dataset must have labels attribute."
self
.
sample_per_label
=
sample_per_id
...
...
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