feat: Add new --footprint-type geojson/csv option to export footpri…

…nts as CSV (#51)

satproc/chips.py

@@ -18,7 +18,12 @@
     prepare_label_shapes,
     write_window_masks,
 )
-from satproc.utils import rescale_intensity, sliding_windows, write_chips_geojson
+from satproc.utils import (
+    rescale_intensity,
+    sliding_windows,
+    write_chips_geojson,
+    write_chips_csv,
+)
 
 __author__ = "Damián Silvani"
 __copyright__ = "Dymaxion Labs"
@@ -40,6 +45,7 @@ def extract_chips(
     bands=None,
     chip_type="tif",
     write_footprints=False,
+    footprints_type="geojson",
     classes=None,
     crs=None,
     skip_existing=True,
@@ -85,6 +91,7 @@ def extract_chips(
                 chip_type=chip_type,
                 within=within,
                 write_footprints=write_footprints,
+                footprints_type=footprints_type,
                 crs=crs,
                 labels=labels,
                 label_property=label_property,
@@ -108,6 +115,7 @@ def extract_chips_from_raster(
     bands=None,
     chip_type="tif",
     write_footprints=False,
+    footprints_type="geojson",
     labels=None,
     label_property="class",
     mask_type="class",
@@ -159,10 +167,12 @@ def extract_chips_from_raster(
         if bands is None:
             bands = list(range(1, min(ds.count, 3) + 1))
 
-        _logger.info((
-            f"Building windows of size {size}, step size {step_size}"
-            f"{' (no overlap)' if step_size >= size else ''}"
-        ))
+        _logger.info(
+            (
+                f"Building windows of size {size}, step size {step_size}"
+                f"{' (no overlap)' if step_size >= size else ''}"
+            )
+        )
         win_size = (size, size)
         win_step_size = (step_size, step_size)
         windows = list(
@@ -277,15 +287,30 @@ def filter_fn(w, aoi):
                 )
 
         if write_footprints:
-            geojson_path = os.path.join(output_dir, "{}.geojson".format(basename))
-            _logger.info("Write chips footprints GeoJSON at %s", geojson_path)
-            write_chips_geojson(
-                geojson_path,
-                chips,
-                chip_type=chip_type,
-                crs=str(meta["crs"]),
-                basename=basename,
-            )
+            if footprints_type == "geojson":
+                if not meta["crs"]:
+                    _logger.error("No CRS found, cannot write GeoJSON footprints")
+                else:
+                    geojson_path = os.path.join(
+                        output_dir, "{}.geojson".format(basename)
+                    )
+                    _logger.info("Write chips footprints GeoJSON at %s", geojson_path)
+                    write_chips_geojson(
+                        geojson_path,
+                        chips,
+                        chip_type=chip_type,
+                        crs=str(meta["crs"]),
+                        basename=basename,
+                    )
+            elif footprints_type == "csv":
+                csv_path = os.path.join(output_dir, "{}.csv".format(basename))
+                _logger.info("Write chips footprints CSV at %s", csv_path)
+                write_chips_csv(
+                    csv_path,
+                    chips,
+                    chip_type=chip_type,
+                    basename=basename,
+                )
 
 
 def write_image(img, path, *, percentiles=None, skip_low_contrast=False):

satproc/console/extract_chips.py

@@ -50,7 +50,10 @@ def parse_args(args):
         "--size", type=int, default=256, help="size of image tiles, in pixels"
     )
     parser.add_argument(
-        "--step-size", type=int, default=None, help="step size (i.e. stride), in pixels (if None, same as size i.e. no overlap)"
+        "--step-size",
+        type=int,
+        default=None,
+        help="step size (i.e. stride), in pixels (if None, same as size i.e. no overlap)",
     )
     parser.add_argument(
         "--sliding-windows-mode",
@@ -165,6 +168,12 @@ def parse_args(args):
         dest="write_footprints",
         action="store_false",
     )
+    parser.add_argument(
+        "--footprints-type",
+        help="footprints file format",
+        choices=["geojson", "csv"],
+        default="geojson",
+    )
 
     parser.add_argument("--crs", help="force CRS of input files")
 
@@ -303,6 +312,7 @@ def main(args):
         chip_type=args.type,
         within=args.within,
         write_footprints=args.write_footprints,
+        footprints_type=args.footprints_type,
         classes=args.classes,
         crs=args.crs,
         skip_existing=args.skip_existing,

satproc/utils.py

@@ -1,3 +1,4 @@
+import csv
 import json
 import logging
 import multiprocessing as mp
@@ -184,6 +185,42 @@ def write_chips_geojson(output_path, chip_pairs, *, chip_type, crs, basename):
         f.write(json.dumps(d))
 
 
+def write_chips_csv(output_path, chip_pairs, *, chip_type, basename):
+    """Write a CSV containing chips polygons as features
+
+    Parameters
+    ----------
+    output_path : str
+        CSV output path
+    chip_pairs : Tuple[Shape, Tuple[int, int, int]]
+        a pair with the chip polygon geometry, and a tuple of (feature id, x, y)
+    chip_type : str
+        chip file type extension (e.g. tif, jpg)
+    basename : str
+        basename of chip files
+
+    Returns
+    -------
+    None
+
+    """
+    if not chip_pairs:
+        _logger.warn("No chips to save")
+        return
+
+    _logger.info("Write chips CSV")
+    os.makedirs(os.path.dirname(output_path), exist_ok=True)
+
+    header = ["id", "x", "y", "minx", "miny", "maxx", "maxy", "filename"]
+    with open(output_path, "w") as f:
+        writer = csv.writer(f)
+        writer.writerow(header)
+        for i, (chip, (_fi, xi, yi)) in enumerate(chip_pairs):
+            minx, miny, maxx, maxy = chip.bounds
+            filename = f"{basename}_{xi}_{yi}.{chip_type}"
+            writer.writerow([i, xi, yi, minx, miny, maxx, maxy, filename])
+
+
 def get_raster_band_count(path):
     """Get raster band count