Deep Learning - "Unknown dataset_type="KITTI_rectangles"

NickPombo

<\/HEAD>

Olá,<\/SPAN><\/P>Atualmente estou trabalhando no treinamento de um modelo de deep learning no ArcGIS Pro que detectará imagens geotermais em imagens Landsat 8. Tenho seguido este tutorial: Train Deep Learning Model94Help | Documentation<\/A> .<\/SPAN><\/P>

<\/P>

Estou atualmente travado na última etapa do processo, treinando o modelo em si. Criei uma pasta adequada para todos os dados de treinamento exportados e agora estou trabalhando com Python. Aqui está meu código abaixo:<\/SPAN><\/P><\/P>importar<\/SPAN> <\/SPAN>arcpy <\/SPAN><\/P>de<\/SPAN> <\/SPAN>arcpy.ia <\/SPAN><\/SPAN>importar<\/SPAN> <\/SPAN>* <\/SPAN><\/P> <\/SPAN><\/P># Verifique a licença da extensão ArcGIS Image Analyst<\/P>arcpy.CheckOutExtension("ImageAnalyst")

#Definir parâmetros de entrada

in_folder = "C:\Users\gphy426\Documents\Geothermal Pombo\training_data_geoproc"

out_folder = "C:\Users\gphy426\Documents\Geothermal Pombo\trained_model"

max_epochs = 20

model_type = "FEATURE_CLASSIFIER"

batch_size = 2

arg = "# #"

learning_rate = None

backbone_model = "RESNET34"

pretrained_model = None

validation_percent = 10

stop_training = "STOP_TRAINING"

 

# Executar<\/p>

TrainDeepLearningModel(in_folder, out_folder, max_epochs, model_type,
                                      bacth_size, arg, learning_rate, backbone_model, pretrained_model,
                        &nbs pvalidation_percent, stop_training)
</p>
<p>
</p>
<p>
The error I seem to be running into now is "unknown dataset_type = KITTI_rectangles".&amp ; nbsp;
I&amp ; apos;m attached an image of the exact error that I am receiving to this forum post. I have done a bit of forum reading to see how I could bypass this issue, but the main solution I&amp ; aposve seen regarding pascal VOC objects does not work for my case, considering that the reference raster is two bands (not the optimal one or three).
I run into the same issue when using the geoprocessing tool "train deep learning model".
Any thoughts or suggestions are appreciated.
</body>
</html>

SangeetMathew

Hi Nick, 

The training api and tools use  "transfer learning" to train our models which means your data needs to be 3 Band 8 bit. What I would suggest to make your dataset training capable, is to use a composite band tool/function on your dataset to make it 3 band( Here is some help How To: Convert individual raster bands into a color or multiband image ). If it is not 8 bit you can use the stretch function to make it 8 bit. Then run the Export training tool (using PASCAL VOC Objects option ) and then train your data. Hope this helps.

Thanks

Sangeet