Merge pull request #9 from patcmorneau/main

Update README.md

README.md

@@ -2,7 +2,6 @@
 Mussel prediction from MBES point cloud
 ## 1) Prep ground truth data
 ### Remove useless columns, sort by date
-consider epsg 8254 to be equal to epsg 4326
 ```
 cut -d ";" -f 1,2,3,5 data/mussels/merge_of_raw_data.csv | sort -t";" -k4 -r > mussels_sorted_by_date_epsg-8254.csv
 ```
@@ -21,82 +20,18 @@ sed -i '1,4d' mussels_sorted_by_date_epsg-8254.csv
 cut -d ";" -f 1-3 mussels_sorted_by_date_epsg-8254.csv | sort -k1,1 -k2,2 --unique > mussels_epsg-8254.csv
 ```
 
-## 2) Prep MBES data
-### Convert MBES dms to decimal
+## 2) Train regression model
 ```
-python3 src/dms_to_dec.py data/test/cap-sample-epsg8254-dms.txt > data/test/cap-sample-epsg8254-decimal.txt
+python3 src/train_model.py data/train_data/training_data.txt
 ```
-
-## 3) Put MBES and ground truth data in same reference system -> ENU
-### Compile 
-```
-cd MBES-lib
-mkdir build && cd build
-cmake ..
-make wgs2lgf
-```
-```
-g++ -I /usr/include/eigen3 src/wgs2lgf_from_lat_lon_eh.cpp -o wgs2lgf_from_lat_lon_eh
-```
-
-keep centroid values somewhere
-```
-cat Cap-Rouge_to_Lac-St-Pierre_epsg-8254-decimal.txt | ./MBES-lib/build/wgs2lgf enu > Cap-Rouge_to_Lac-St-Pierre_enu.txt
-```
-lat lon Z centroid :  46.527207203821909332 -72.066496911800527414 12.819235715989787394
-```
-cat mussels_epsg-8254.txt | ./wgs2lgf_from_lat_lon_groundruth enu centroid_Lat centroid_Lon centroid_ellipsoidal_height > mussels_enu.txt
-```
-
-### Generate hackel features
-```
-cat ~/Cap-Rouge_to_Lac-St-Pierre_enu.txt | ./soundings_generate_features 10 > ~/Cap-Rouge_to_Lac-St-Pierre_enu.hackel
-```
-
-### Unsupervised MBES classification
-```
-python3 gmm_best_fit.py ~/Cap-Rouge_to_Lac-St-Pierre_enu.hackel 6 > Cap-Rouge_to_Lac-St-Pierre_enu_gmm_hackel.txt
-```
-
-## 4) Generate Trainning data
-
-output xyz hackel_features gmmClass musselGroundTruth
+## 3) Generate hackel file
+### Mbes file must be in a projected or orthogonal reference system on a 1*1 M grid
+### The utility soundings_generate_features is from Benthinc_classifier : https://github.com/CIDCO-dev/BenthicClassifier
 ```
-python3 generate_training_data.py ../data/mussels/mussels_enu_more_precision_centroid.txt ~/Cap-Rouge_to_Lac-St-Pierre_enu_hackel_gmm.txt 10  > ~/training_data.txt
-```
-
-## 5) Optimize model parameter and train
-all cpu core will be use for parameter optimisation
-```
-python3 train_model.py training_data.txt
+cat MBES_FILE.txt | ./soundings_generate_features 10 > outputFilePath.hackel
 ```
 
 ## 6) Apply model
 ```
-python3 apply_model.py trained_mussel_regression.model FILE > ~/result.xyzc
-```
-
-## 7) Rasterize result
-### Compile utility
-```
-g++ -I /usr/include/eigen3 lgf2wgs.cpp -o lgf2wgs
-```
-
-### convert coordinates back to WGS
-```
-cat ~/result.xyzc | ./lgf2wgs enu 46.527207203821909332 -72.066496911800527414 12.819235715989787394 > ~/lat_lon_eh_Class.csv
-```
-
-### get min max xy to change raster resolution
-```
-python3 find_min_max_coordinates.py ~/lat_lon_eh_Class.csv
-```
-
-### Export result as geotiff
-```
-bash script/rasterize.bash ~/lat_lon_eh_Class.csv ~/rasters 4326
-```
-or run a similar command like this
-```
-gdal_grid -zfield "field_3" -a_srs EPSG:4326 -a invdist -ot Float64 -l lat_lon_eh_Class script/lat_lon_eh_Class.vrt ~/raster_reso_1/mussel_resolution_1.tiff --config GDAL_NUM_THREADS ALL_CPUS -txe -72.6806747039285 -71.41272166174498 -tye 46.26391417288448 46.71285611583465 -tr 0.00001 0.00001
+python3 apply_model.py trained_mussel_regression.model FILE.hackel > outputFilePath_result.csv
 ```