spacenetchallenge / buildingdetectors_round2 Goto Github PK

Hi all,

While trying to train the solution I am facing the following issue . Any one else faced this ?
any solution or hints I am stuck at this issue.

I am using coda 10.

19-11-07 12:17:35,651 INFO Preproc for training on AOI_3_Paris ... done
python v16.py preproc_train /data/train/AOI_3_Paris_Train/
Using Theano backend.
WARNING (theano.sandbox.cuda): The cuda backend is deprecated and will be removed in the next release (v0.10). Please switch to the gpuarray backend. You can get more information about how to switch at this URL:
https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29

Using gpu device 0: GeForce RTX 2080 Ti (CNMeM is disabled, cuDNN 5110)
Traceback (most recent call last):
File "v16.py", line 36, in
from keras.models import Model
File "/opt/conda/envs/py35/lib/python3.5/site-packages/keras/init.py", line 2, in
from . import backend
File "/opt/conda/envs/py35/lib/python3.5/site-packages/keras/backend/init.py", line 61, in
from .theano_backend import *
File "/opt/conda/envs/py35/lib/python3.5/site-packages/keras/backend/theano_backend.py", line 1, in
import theano
File "/opt/conda/envs/py35/lib/python3.5/site-packages/theano/init.py", line 116, in
theano.sandbox.cuda.tests.test_driver.test_nvidia_driver1()
File "/opt/conda/envs/py35/lib/python3.5/site-packages/theano/sandbox/cuda/tests/test_driver.py", line 41, in test_nvidia_driver1
raise Exception("The nvidia driver version installed with this OS "
Exception: The nvidia driver version installed with this OS does not give good results for reduction.Installing the nvidia driver available on the same download page as the cuda package will fix the problem: http://developer.nvidia.com/cuda-downloads

TRAINING v9s model
python v9s.py validate /data/train/AOI_3_Paris_Train/
Using Theano backend.
WARNING (theano.sandbox.cuda): The cuda backend is deprecated and will be removed in the next release (v0.10). Please switch to the gpuarray backend. You can get more information about how to switch at this URL:
https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29

In Nvidia they are suggesting to stop using Theano itself.
https://devtalk.nvidia.com/default/topic/1056698/cuda-setup-and-installation/exception-the-nvidia-driver-version-installed-with-this-os-does-not-give-good-results-for-reduction/
in google there is no proper solution is there.

Hi,

Is there more documentation on how to run this? I'm having trouble with training from scratch. Also, is "kohei-solution-20170612_4.zip" available somewhere?

Thanks

Hello @dlindenbaum,
In 2-wleite solution, while training I am facing the following issue
OutOfMemoryError: Java heap space

What should be CPU and RAM configuration for this solution? I am running at windows OS , Processor AMD Ryzen 5 3600 6-core 3.59GHz
And RAM 16.0 GB
any solution or hints

Thank you for uploading this. The results all look very neat. However when I'm trying to replicate the results, I'm having trouble downloading the dataset to local disk and really don't know how to resolve this problem. Any help is sincerely appreciated.

I excute this from command line:

proxychains aws s3 cp s3://spacenet-dataset/spacenet/SN2_buildings/tarballs/SN2_buildings_train_AOI_5_Khartoum.tar.gz . --no-sign-request

It'll download just fine for a couple of minutes and then break (completed about 3.8G of this 4.7G file), indicating Errno 22 with Invalid Argument. I've also tried all other three trainning dataset but all encountered the same problem. I succeeded using the same command to download full AOI 3 – Paris – Building Extraction Testing (1.9G) and AOI 5 – Khartoum – Building Extraction Testing (1.6G).

download failed: s3://spacenet-dataset/spacenet/SN2_buildings/tarballs/SN2_buildings_train_AOI_5_Khartoum.tar.gz to ./SN2_buildings_train_AOI_5_Khartoum.tar.gz [Errno 22] Invalid argument

And this is my aws cli version etc.

aws-cli/1.18.69 Python/3.6.9 Linux/5.4.0-51-generic botocore/1.16.19

Hi, I find the link to download osm data is not available now. Is there another link to download them?

hi，I am Newby to DeepLearning。 I can not download the instruction of XD_XD，so I do not how to use the code and imagery data downloading from the AWS,. I am looking forward to your response

Hello @dlindenbaum, According to band-triplets.txt Dataset have 8 bands images.
My question is , Can we convert those 8 Band images to 4 bands including RGB+NIR images? As shown in attached photo.
If yes , please guide how can i read RGB and NIR channel and merge them to save them as an image.

I've trained both of these models (v9s and v13) and the outputs from evalfscore are bad. The polygons are too large and encompass multiple buildings.
Any tips?