README instructions do not work

[gaetanbahl]

• Instead of cp -r ./ese_seg/gluon-cv gluon-cv it should be something like cp -r ./ese_seg/gluon-cv/* gluon-cv/.
• After tar -zxvf VOCsbdche.tar.gz we need to do mv new_VOCsbdche VOCsbdche because the path is hardcoded
• It need to be specified that mxnet is not enough. mxnet-cu100 (with cuda support) is needed.
• /home/whymit/ESE-SEG/data is hardcoded into the dataset class and there is no argument to change it. Need to add root=<your path to data> to line 102 and 108 in sbd_train_che_8.py

Hope that helps anyone trying to run this code.

[curryJ]

Thanks for your help But I think the the path of gluoncv is not correct.When I run this code 'from gluoncv.utils.metrics.voc_polygon_detection import VOC07PolygonMApMetric',thers is an error that no module named VOC07PolygonMApMetric.You can visite my issue https://github.com/WenqiangX/ese_seg/issues/10. I solve this by copying the full directory to my python site-packages path. Do you run this code with your own dataset? Does it success? I train on my own dataset but I get this issue https://github.com/WenqiangX/ese_seg/issues/11 Thank you.

[curryJ]

By the way,I install gluoncv through pip--> pip install gluoncv

[Kismetzc]

* Instead of `cp -r ./ese_seg/gluon-cv gluon-cv`  it should be something like `cp -r ./ese_seg/gluon-cv/* gluon-cv/`.

* After `tar -zxvf VOCsbdche.tar.gz` we need to do `mv new_VOCsbdche VOCsbdche` because the path is hardcoded

* It need to be specified that `mxnet` is not enough. `mxnet-cu100` (with cuda support) is needed.

* `/home/whymit/ESE-SEG/data` is hardcoded into the dataset class and there is no argument to change it. Need to add `root=<your path to data>` to line 102 and 108 in sbd_train_che_8.py

Hope that helps anyone trying to run this code.

Thanks! It works.