From: @cathwong
Reviewed-by: 
Signed-off-by:

for image/vision datasets

From: @zjun3021
Reviewed-by: 
Signed-off-by:

From: @luoyang42
Reviewed-by: 
Signed-off-by:

From: @zhangbuxue
Reviewed-by: @ginfung
Signed-off-by:

From: @liangzhibo
Reviewed-by: @kingxian,@jpc_chenjianping
Signed-off-by: @kingxian

From: @zhangbuxue
Reviewed-by: 
Signed-off-by:

From: @ziruiwu
Reviewed-by: 
Signed-off-by:

From: @alexyuyue
Reviewed-by: @robingrosman,@nsyca
Signed-off-by: @robingrosman

From: @robingrosman
Reviewed-by: 
Signed-off-by:

Signed-off-by: alex-yuyue <yue.yu1@huawei.com>

Signed-off-by: zjun <zhangjun0@huawei.com>

Modify real dynamic

Signed-off-by: zjun <zhangjun0@huawei.com>

From: @yepei6
Reviewed-by: 
Signed-off-by:

From: @lizhenglong1992
Reviewed-by: @pandoublefeng,@heleiwang
Signed-off-by: @pandoublefeng

Adjust memory usage to not increase as parallelism increases.  it will
stay at same level it would be with 4 parallelism

From: @shenwei41
Reviewed-by: 
Signed-off-by:

From: @cathwong
Reviewed-by: @robingrosman
Signed-off-by: @robingrosman

From: @liu_xiao_93
Reviewed-by: @liangchenghui
Signed-off-by: @liangchenghui

From: @liubuyu
Reviewed-by: 
Signed-off-by:

From: @ezphlow
Reviewed-by: 
Signed-off-by:

From: @rainyhorse
Reviewed-by: @xsmq,@zhoufeng54
Signed-off-by: @zhoufeng54

what if we just do nothing

trying wrapper approach

Fix yolo

From: @mahdirahmanihanzaki
Reviewed-by: 
Signed-off-by:

From: @lianliguang
Reviewed-by: 
Signed-off-by:

From: @zhupuxu
Reviewed-by: 
Signed-off-by:

From: @lijiaqi0612
Reviewed-by: 
Signed-off-by:

Signed-off-by: zhupuxu <zhupuxu@huawei.com>

From: @wanyiming
Reviewed-by: 
Signed-off-by:

From: @lizhenglong1992
Reviewed-by: @liucunwei
Signed-off-by: @liucunwei

From: @hwhewei
Reviewed-by: 
Signed-off-by: