Deep Learning - "Neznámý dataset_type="KITTI_rectangles"

NickPombo

<\/HEAD>

Dobrý den,<\/SPAN><\/P>Momentálně pracuji na tréninku modelu deep learning v ArcGIS Pro, který bude detekovat geotermální snímky v Landsat 8 snímcích. Řídím se tímto tutoriálem: Train Deep Learning ModelHelp | Dokumentace<\/A> .<\/SPAN><\/P>

<\/P>

Momentálně jsem zaseknutý na posledním kroku procesu, samotném tréninku modelu. Vytvořil jsem správnou složku pro všechna exportovaná tréninková data a nyní pracuji s Pythonem. Zde je můj kód níže:<\/SPAN><\/P><\/P>importujte<\/SPAN> <\/SPAN>arcpy <\/SPAN><\/P>z<\/SPAN> <\/SPAN>arcpy.ia <\/SPAN><\/SPAN>importujte<\/SPAN> <\/SPAN>* <\/SPAN><\/P> <\/SPAN><\/P># Zkontrolujte licenci rozšíření ArcGIS Image Analyst<\/P>arcpy.CheckOutExtension("ImageAnalyst"<\/SPAN>)<\/P>
<\/P># Definujte vstupní parametry<\/P>in_folder = "C:\\Users\\gphy426\\Documents\\Geothermal Pombo\\training_data_geoproc"<\/SPAN>
out_folder = "C:\\Users\\gphy426\\Documents\\Geothermal Pombo\\trained_model"<\/SPAN>
max_epochs = 20<\/SPAN>
model_type = "FEATURE_CLASSIFIER"<\/SPAN>
batch_size = 2<\/SPAN>
arg = "# #"<\/SPAN>
learning_rate = None<\/SPAN>
backbone_model = "RESNET34"<\/SPAN>
pretrained_model = None<\/SPAN>
validation_percent = 10<\/SPAN>
stop_training = "STOP_TRAINING"<\/SPAN>

# Spusťte příkaz<\/P>

TrainDeepLearningModel(in_folder, out_folder, max_epochs, model_type,
                                 bath_size, arg, learning_rate, backbone_model, pretrained_model,
                             &nbpspvalidation_percent, stop_training)

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