Example optimisation fails

[cre...@gmail.com]

Hi João,

running your example data (and also my own data) with your tutorial code for parameter optimisation I am allways running in the below error message. Obvoiusly it is thrown by the GA package. As far as I understand all parameters are correctly passed through. Any suggestions are welcome.
best Chris


rstFeatures <- file.path(envrmt$path_segoptim,"SegmFeat_WV2_b532.tif")
trainData <- file.path(envrmt$path_segoptim,"TrainAreas.tif")

# 3 - start code 
#-----------------

pyPath <- "C:/Anaconda3/envs/py35"
# Minimum values for nr of clusters, min segment size and spectral threshold
RSGISLib_Shep_min <- c(5, 10, 3)

# Maximum values for nr of clusters, min segment size and spectral threshold
RSGISLib_Shep_max <- c(12, 60, 40)

# Run the optimization procedure ----------------------------------------------------------------------------- #


gaOptim <- gaOptimizeSegmentationParams(  rstFeatures = rstFeatures,
                                          trainData   = trainData,
                                          # // Segmentation parameters ---
                                          segmentMethod = "RSGISLib_Shep",
                                          inputRstPath  = envrmt$path_segoptim,
                                          pythonPath    = pyPath, 
                                          verbose = FALSE,
                                          # // End segmentation parameters ---
                                          trainThresh   = 0.5,
                                          segmStatsFuns = c("mean"),
                                          classificationMethod = "RF",
                                          classificationMethodParams = NULL,
                                          balanceTrainData = FALSE,
                                          balanceMethod = "ubOver",
                                          evalMethod = "5FCV",
                                          evalMetric = "Kappa",
                                          minTrainCases = 30,
                                          minCasesByClassTrain = 10,
                                          minCasesByClassTest = 10,
                                          minImgSegm = 30,
                                          lower = RSGISLib_Shep_min,
                                          upper = RSGISLib_Shep_max,
                                          popSize = 20, 
                                          pcrossover = 0.8, 
                                          pmutation = 0.2,
                                          maxiter = 100, 
                                          run = 20,
                                          keepBest = TRUE,
                                          parallel = 2) # us


GA | iter = 1 | Mean = NaN | Best = -Inf
GA | iter = 2 | Mean = NaN | Best = -Inf
Error in if (object@run >= run) break : 
 Missing value, where TRUE/FALSE is necessary
Additionally: Warning messages:
1: In max(fitness) :
 no non-missing argument for max; return -Inf
2: In max(fitness, na.rm = TRUE) :
 no non-missing argument for max; return -Inf
3: In max(fitness, na.rm = TRUE) :
 no non-missing argument for max; return -Inf
4: In max(fitness) :
 no non-missing argument for max; return -Inf
5: In max(fitness, na.rm = TRUE) :
 no non-missing argument for max; return -Inf
6: In max(x, na.rm = TRUE) :
 no non-missing argument for max; return -Inf
Error in gaOptimizeSegmentationParams(rstFeatures = rstFeatures, trainData = trainData, : 
 An error occurred while running ga function!

sessionInfo()
R version 3.6.1 (2019-07-05)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 18363)

Matrix products: default

locale:
[1] LC_COLLATE=German_Germany.1252  LC_CTYPE=German_Germany.1252    LC_MONETARY=German_Germany.1252
[4] LC_NUMERIC=C                    LC_TIME=German_Germany.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] SegOptim_0.2.3                    link2GI_0.4.3                     envimaR_0.1.0                    
 [4] SuperpixelImageSegmentation_1.0.1 OpenImageR_1.1.6                  hrbrthemes_0.6.0                 
 [7] ggrepel_0.8.1                     scico_1.1.0                       ggnewscale_0.4.1                 
[10] ggplot2_3.2.1                     dplyr_0.8.4                       here_0.1                         
[13] rnaturalearth_0.1.0               rgdal_1.4-8                       sp_1.4-1                         
[16] stars_0.4-0                       sf_0.8-1                          abind_1.4-5                      

loaded via a namespace (and not attached):
  [1] mlr_2.17.0           backports_1.1.5      fastmatch_1.1-0      systemfonts_0.1.1    lwgeom_0.2-1        
  [6] plyr_1.8.5           lazyeval_0.2.2       splines_3.6.1        crosstalk_1.0.0      leaflet_2.0.3       
 [11] usethis_1.5.1        digest_0.6.25        foreach_1.4.8        htmltools_0.4.0      tiff_0.1-5          
 [16] gdata_2.18.0         fansi_0.4.1          magrittr_1.5         checkmate_2.0.0      memoise_1.1.0       
 [21] BBmisc_1.11          unbalanced_2.0       cluster_2.1.0        doParallel_1.0.15    ROCR_1.0-7          
 [26] remotes_2.1.1        recipes_0.1.9        gower_0.2.1          extrafont_0.17       extrafontdb_1.0     
 [31] prettyunits_1.1.1    jpeg_0.1-8.1         colorspace_1.4-1     xfun_0.12            callr_3.4.2         
 [36] crayon_1.3.4         roxygen2_7.0.2       survival_2.44-1.1    iterators_1.0.12     glue_1.3.1          
 [41] gtable_0.3.0         cclust_0.6-21        ipred_0.9-9          webshot_0.5.2        pkgbuild_1.0.6      
 [46] Rttf2pt1_1.3.8       scales_1.1.0         uavRst_0.5-5         DBI_1.1.0            rngtools_1.5        
 [51] Rcpp_1.0.3           viridisLite_0.3.0    xtable_1.8-4         units_0.6-5          stats4_3.6.1        
 [56] lava_1.6.6           prodlim_2019.11.13   htmlwidgets_1.5.1    clusterCrit_1.2.8    FNN_1.1.3           
 [61] gplots_3.0.3         ellipsis_0.3.0       pkgconfig_2.0.3      ParamHelpers_1.13    nnet_7.3-12         
 [66] caret_6.0-85         reshape2_1.4.3       tidyselect_1.0.0     rlang_0.4.4          later_1.0.0         
 [71] munsell_0.5.0        tools_3.6.1          cli_2.0.2            generics_0.0.2       devtools_2.2.2      
 [76] evaluate_0.14        stringr_1.4.0        fastmap_1.0.1        ModelMetrics_1.2.2.1 processx_3.4.2      
 [81] knitr_1.28           fs_1.3.1             caTools_1.18.0       satellite_1.0.2      purrr_0.3.3         
 [86] randomForest_4.6-14  RANN_2.6.1           nlme_3.1-140         doRNG_1.8.2          mime_0.9            
 [91] GA_3.2               xml2_1.2.2           compiler_3.6.1       rstudioapi_0.11      curl_4.3            
 [96] png_0.1-7            e1071_1.7-3          testthat_2.3.1       tibble_2.1.3         stringi_1.4.6       
[101] ps_1.3.2             desc_1.2.0           gdtools_0.2.1        lattice_0.20-38      Matrix_1.2-17       
[106] classInt_0.4-2       gbm_2.1.5            pillar_1.4.3         lifecycle_0.1.0      data.table_1.12.8   
[111] bitops_1.0-6         mapview_2.7.0        raster_3.0-12        httpuv_1.5.2         R6_2.4.1            
[116] promises_1.1.0       KernSmooth_2.23-15   gridExtra_2.3        sessioninfo_1.1.1    codetools_0.2-16    
[121] MASS_7.3-51.4        gtools_3.8.1         assertthat_0.2.1     pkgload_1.0.2        rprojroot_1.3-2     
[126] withr_2.1.2          parallel_3.6.1       grid_3.6.1           rpart_4.1-15         timeDate_3043.102   
[131] class_7.3-15         rmarkdown_2.1        mda_0.4-10           parallelMap_1.4      pROC_1.16.1         
[136] base64enc_0.1-3      lubridate_1.7.4      shiny_1.4.0 

[cre...@gmail.com]

In addition:

after diving in it seems that the NA pixel makes some trouble (should*nt?!) But after remowving NA data from the training data file an error is thrown by the GA function at row 228:

               { fit <- do.call(fitness, c(list(Pop[i,]), callArgs))

SpectralRange=0.18225
SpatialRange=16.92107
MinSize=811.43608
Error in .local(.Object, ...) : 

Error in .rasterObjectFromFile(x, band = band, objecttype = "RasterLayer",  : 
  Cannot create a RasterLayer object from this file. (file does not exist)

Actually the above used

rstFeatures = rstFeatures,
inputRstPath  = envrmt$path_segoptim,

cheers Chris
are identical and point to the tif file.

[João Gonçalves]

Hi Chris,

I am guessing that you are still using the data in the tutorial and in that case NA pixels are not a problem. These are removed prior to the supervised classification step.

As for RSGISLib please make sure that you are using version 3.5 (or lower). Check out the installation instructions here. You can also test other segmenters such as OTB or SAGA which usually have less issues and are compatible to their latest versions. If you do this, please remind to change the min/max parameters accordingly.

Also confirm that you have full read/write permissions to your target output folder, in this case your working directory in R. The issue that you describe is often related to this. The segmentation software will use this folder to place temp files from the segmentation step and it seems to be failing. In windows systems like yours, opening R/RStudio with 'administrator permissions' usually works better.

Finally, for making a better diagnose of your situation please run the optimization firstly with parallel = FALSE which makes it easier to spot errors/issues. Please report the messages here for further inspection.

Thanks for your patient and for using SegOptim.

Cheers

- - -

[cre...@gmail.com]

Hi João.

all good things go by three - so sorry for bothering you.

Finally I managed it to identify the error. It was a code table issue from transferring script data from Linux to Win10 therefore whysoever the paths to the binaries were misinterpreted
and neither RSGISlib nor OTB binaries were found. The resulting error message was: "Cannot create a RasterLayer object from this file. (file does not exist)"...

So it's time to say thanks for the cool package!
cheers Chris

[cre...@gmail.com]

Hi João,
this posts were in parallel... Yes I have had checked your hints already. I finally found it debugging thru your package and checking the OTB wrapper commands and there was something like a hidden blank  in the string. Acutally I don't now why. On this way I noticed that I should change the NAs too. So everything fine for now.
Thanks again!
Cheers

[João Gonçalves]

Thanks for the feedback Chris!

Let me know if you have any further question, bug report or feature request.

Cheers

- - -