ANN for Geothermal classification
python create_doe_dataset.py -i ../doe-som/brady_som_output.gri -c 3 -d brady_samples_19x3d -s 100000 -k 19
- Create a dataset using the GRI file as source for the rasters. Creates tiles with 3 channels (bands),and the output is located in the directory brady_samples_19x3d.
- It creates 100,000 samples and the kernel dimensions will be 19x19
python doe_ann3.py -a -d brady_samples_19x3d -l tmp/doe_19x3d200b.l -m tmp/doe19x3d200b.h5 -k 19 -b 200 -e 200 -c 3 -r
- Creates from scratch (-r : reset), and with data augmentation (-a), an AI with samples from the directory brady_samples_19x3d. It will output a lables file doe_19x3d200b.l and a model file doe19x3d200b.h5 (this will contain the network architecture and all weights).
- The kernel to use is 19x19, it will run batches of size 200 (-b: batch size) and will runfor 200 full iterations (-e: epochs). The Netwrok will use 3 data bands."
python doe_ann_map.py -i ../doe-som/brady_som_output.gri -k 19 -m tmp/doe_19x3d200b.h5 -l tmp/doe_19x3d200b.l
- Creates a map using as basis the raster from image (-i) brady_som_output.gri. The kernel to us is size 19x19, and the model and labels are in the files doe_19x3d200b.h5 and doe_19x3d200b.l