Make sure training resumes for restarted jobs

[lewfish]

This may be a factor #106

[jamesmcclain]

As I understand it, the task is to verify that this is now the case. Should it be true for the feature branch, or is there another branch that I should use?

[lossyrob]

This is now in the feature/api-refactor branch

[jamesmcclain]

This is now in the feature/api-refactor branch

Okay, sounds good.

[jamesmcclain]

Chip Classification

rastervision run aws_batch -e spacenet.chip_classification -a root_uri ${RVROOT}

then

rastervision run aws_batch -e spacenet.chip_classification -a root_uri ${RVROOT} --rerun train

Confirmed working:

[jamesmcclain]

Object Detection

export ROOT_URI='s3://raster-vision-mcclain/xview'
export DATA_URI='s3://raster-vision-xview'
rastervision run aws_batch -e xview.object_detection -f *mobilenet* -a root_uri ${ROOT_URI} -a data_uri ${DATA_URI}

then

rastervision run aws_batch -e xview.object_detection -f *mobilenet* -a root_uri ${ROOT_URI} -a data_uri --rerun train`

Confirmed working.

[jamesmcclain]

Segmentation

export ROOT_URI='s3://raster-vision-mcclain/SEGMENTATION'
export DATA_URI='s3://raster-vision-raw-data/isprs-potsdam'
rastervision run aws_batch -e potsdam.semantic_segmentation -a root_uri ${ROOT_URI} -a data_uri ${DATA_URI}

then

rastervision run aws_batch -e potsdam.semantic_segmentation -a root_uri ${ROOT_URI} -a data_uri ${DATA_URI} --rerun train

Confirmed working.