From ad70512b3e40b65788b775c7560335d300ff02ac Mon Sep 17 00:00:00 2001
From: Sam Gillingham <gillingham.sam@gmail.com>
Date: Thu, 4 Jul 2024 22:48:08 +0000
Subject: [PATCH] fixes and tidy

---
 .../awsbatch/template/template.yaml              |  4 ++--
 pyshepseg/tilingstats.py                         | 16 ++++++++++------
 2 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/parallel_examples/awsbatch/template/template.yaml b/parallel_examples/awsbatch/template/template.yaml
index 92d8873..50e9f78 100644
--- a/parallel_examples/awsbatch/template/template.yaml
+++ b/parallel_examples/awsbatch/template/template.yaml
@@ -190,8 +190,8 @@ Resources:
       JobDefinitionName: PyShepSegBatchJobDefinitionStitch
       ContainerProperties:
         Image: !Join ['', [!GetAtt BatchRepository.RepositoryUri, ":latest"]]
-        Vcpus: 2
-        Memory: 24000
+        Vcpus: 4
+        Memory: 12000
       RetryStrategy:
         Attempts: 1
   BatchProcessingJobQueue:
diff --git a/pyshepseg/tilingstats.py b/pyshepseg/tilingstats.py
index 5910e1e..72f867b 100644
--- a/pyshepseg/tilingstats.py
+++ b/pyshepseg/tilingstats.py
@@ -286,7 +286,9 @@ def calcPerSegmentStatsTiledRIOS(imgfile, imgbandnum, segfile,
     del imgds
 
     controls = applier.ApplierControls()
-    controls.selectInputImageLayers([imgbandnum], imgfile)
+    controls.selectInputImageLayers([imgbandnum], 'imgfile')
+    # RIOS default is 256x256. This leads to too many incomplete
+    # segments and increases memory use dramatically. 
     controls.setWindowSize(tiling.TILESIZE, tiling.TILESIZE)
     
     # now create a new temporary file for saving the new columns too
@@ -329,8 +331,9 @@ def calcPerSegmentStatsTiledRIOS(imgfile, imgbandnum, segfile,
     otherArgs.numIntCols = numIntCols
     otherArgs.numFloatCols = numFloatCols
         
-    applier.apply(rioscalcPerSegmentStatsTiled, inputs, outputs, 
+    rtn = applier.apply(rioscalcPerSegmentStatsTiled, inputs, outputs, 
         controls=controls, otherArgs=otherArgs)
+    print(rtn.timings.formatReport())
         
     del tempKEAAttrTbl
     del tempKEABand
@@ -341,7 +344,6 @@ def calcPerSegmentStatsTiledRIOS(imgfile, imgbandnum, segfile,
         raise PyShepSegStatsError('Not all pixels found during processing')
         
     # now merge the stats from the tempfile band info segfile
-    print('copying RAT')
     ratapplier.copyRAT(tempKEA, segfile)
 
 
@@ -1418,7 +1420,9 @@ def calcPerSegmentSpatialStatsTiledRIOS(imgfile, imgbandnum, segfile,
     del imgds
 
     controls = applier.ApplierControls()
-    controls.selectInputImageLayers([imgbandnum], imgfile)
+    controls.selectInputImageLayers([imgbandnum], 'imgfile')
+    # RIOS default is 256x256. This leads to too many incomplete
+    # segments and increases memory use dramatically. 
     controls.setWindowSize(tiling.TILESIZE, tiling.TILESIZE)
     
     # now create a new temporary file for saving the new columns too
@@ -1463,8 +1467,9 @@ def calcPerSegmentSpatialStatsTiledRIOS(imgfile, imgbandnum, segfile,
     otherArgs.userFunc = userFunc
     otherArgs.userParam = userParam
 
-    applier.apply(rioscalcPerSegmentSpatialStatsTiled, inputs, outputs, 
+    rtn = applier.apply(rioscalcPerSegmentSpatialStatsTiled, inputs, outputs, 
         controls=controls, otherArgs=otherArgs)
+    print(rtn.timings.formatReport())
         
     del tempKEAAttrTbl
     del tempKEABand
@@ -1475,7 +1480,6 @@ def calcPerSegmentSpatialStatsTiledRIOS(imgfile, imgbandnum, segfile,
         raise PyShepSegStatsError('Not all pixels found during processing')
 
     # now merge the stats from the tempfile band info segfile
-    print('copying RAT')
     ratapplier.copyRAT(tempKEA, segfile)