Fast AOI polygon query for matching mosaic dataset footprints

Question

Is there any more efficient/faster way to query a mosaic dataset raster catalog (footprints) for records matching an AOI polygon?

This implementation gets slower the larger the AOI polygon is.

import arcpy
import timeit

fgdb_path = 'H:/MY/FGDB/Rasters.gdb'
md_name = 'CIB1'
md_path = fgdb_path + '/' + md_name
md_cat_path = fgdb_path + '/AMD_' + md_name + '_CAT'
aoi_fc = 'H:/MY/FGDB.gdb/AOI'
out_paths = r'in_memory\paths'

# Get AOI poly from FC
with arcpy.da.SearchCursor(aoi_fc,'SHAPE@') as sCur:
     for row in sCur:
         aoi_poly = row[0]

start_time = timeit.default_timer()

# Create OID list of MD raster footprints that the AOI overlaps/contains
oid_list = []
with arcpy.da.SearchCursor(md_path,['SHAPE@', 'OID@']) as sCur:
     for row in sCur:
         if aoi_poly.overlaps(row[0]) or aoi_poly.contains(row[0]):
             oid_list.append(row[1])

elapsed = timeit.default_timer() - start_time
print("Completed in " + str(elapsed))

System Info:

• ArcPy w/ Pro 2.8 python3
• ArcPy w/ ArcGIS Server 10.9 python3

Metrics:

• Mosaic Dataset Footprint: 280K
• Footprint size: 1SqMi
• Small AOI footprint match: 10K results, 30s completion
• Large AOI footprint match: 230K results, 230s completion

Answer 1

I tried on a Mosaic Dataset stored on an external hard-drive (40k rasters). This seems fastest:

Followed by:

aSet = set(row[0] for row in arcpy.da.TableToNumPyArray(r"NORTHLAND_MosaicLayer\Footprint","OID@"))

Note my AOI layer made up from 2945 polygons made of 1779853 (1.8M!) points.