Add multiple versions of the So2Sat dataset

tests/conf/so2sat_all.yaml

@@ -13,4 +13,5 @@ datamodule:
   root: "tests/data/so2sat"
   batch_size: 1
   num_workers: 0
+  version: "2"
   band_set: "all"

tests/conf/so2sat_rgb.yaml

@@ -0,0 +1,18 @@
+module:
+  _target_: torchgeo.trainers.ClassificationTask
+  loss: "ce"
+  model: "resnet18"
+  learning_rate: 1e-3
+  learning_rate_schedule_patience: 6
+  weights: null
+  in_channels: 3
+  num_classes: 17
+
+datamodule:
+  _target_: torchgeo.datamodules.So2SatDataModule
+  root: "tests/data/so2sat"
+  batch_size: 1
+  num_workers: 0
+  version: "3_random"
+  band_set: "rgb"
+  validation_pct: 0.5

tests/conf/so2sat_s1.yaml

@@ -13,4 +13,5 @@ datamodule:
   root: "tests/data/so2sat"
   batch_size: 1
   num_workers: 0
+  version: "2"
   band_set: "s1"

tests/conf/so2sat_s2.yaml

@@ -13,4 +13,5 @@ datamodule:
   root: "tests/data/so2sat"
   batch_size: 1
   num_workers: 0
+  version: "2"
   band_set: "s2"

tests/data/so2sat/data.py

@@ -5,13 +5,14 @@
 
 import hashlib
 import os
+import shutil
 
 import h5py
 import numpy as np
 
-SIZE = 64  # image width/height
+SIZE = 32  # image width/height
 NUM_CLASSES = 17
-NUM_SAMPLES = 1
+NUM_SAMPLES = 2
 
 np.random.seed(0)
 
@@ -28,16 +29,21 @@
     ]
 
     # Random images
-    sen1 = np.random.randint(256, size=(NUM_SAMPLES, SIZE, SIZE, 8), dtype=np.uint8)
-    sen2 = np.random.randint(256, size=(NUM_SAMPLES, SIZE, SIZE, 10), dtype=np.uint8)
+    sen1 = np.random.randint(2, size=(NUM_SAMPLES, SIZE, SIZE, 8), dtype=np.uint8)
+    sen2 = np.random.randint(2, size=(NUM_SAMPLES, SIZE, SIZE, 10), dtype=np.uint8)
 
     # Create datasets
     with h5py.File(filename, "w") as f:
-        f.create_dataset("label", data=label)
-        f.create_dataset("sen1", data=sen1)
-        f.create_dataset("sen2", data=sen2)
+        f.create_dataset("label", data=label, compression="gzip", compression_opts=9)
+        f.create_dataset("sen1", data=sen1, compression="gzip", compression_opts=9)
+        f.create_dataset("sen2", data=sen2, compression="gzip", compression_opts=9)
 
     # Compute checksums
     with open(filename, "rb") as f:
         md5 = hashlib.md5(f.read()).hexdigest()
         print(repr(split.replace("ing", "")) + ":", repr(md5) + ",")
+
+for version in ["random", "block", "culture_10"]:
+    os.makedirs(version, exist_ok=True)
+    shutil.copyfile("training.h5", os.path.join(version, "training.h5"))
+    shutil.copyfile("testing.h5", os.path.join(version, "testing.h5"))

tests/datasets/test_so2sat.py

@@ -21,13 +21,15 @@
 class TestSo2Sat:
     @pytest.fixture(params=["train", "validation", "test"])
     def dataset(self, monkeypatch: MonkeyPatch, request: SubRequest) -> So2Sat:
-        md5s = {
-            "train": "82e0f2d51766b89cb905dbaf8275eb5b",
-            "validation": "bf292ae4737c1698b1a3c6f5e742e0e1",
-            "test": "9a3bbe181b038d4e51f122c4be3c569e",
+        md5s_by_version = {
+            "2": {
+                "train": "56e6fa0edb25b065124a3113372f76e5",
+                "validation": "940c95a737bd2fcdcc46c9a52b31424d",
+                "test": "e97a6746aadc731a1854097f32ab1755",
+            }
         }
 
-        monkeypatch.setattr(So2Sat, "md5s", md5s)
+        monkeypatch.setattr(So2Sat, "md5s_by_version", md5s_by_version)
         root = os.path.join("tests", "data", "so2sat")
         split = request.param
         transforms = nn.Identity()
@@ -51,13 +53,13 @@ def test_getitem(self, dataset: So2Sat) -> None:
         assert isinstance(x["label"], torch.Tensor)
 
     def test_len(self, dataset: So2Sat) -> None:
-        assert len(dataset) == 1
+        assert len(dataset) == 2
 
     def test_out_of_bounds(self, dataset: So2Sat) -> None:
         # h5py at version 2.10.0 raises a ValueError instead of an IndexError so we
         # check for both here
         with pytest.raises((IndexError, ValueError)):
-            dataset[1]
+            dataset[2]
 
     def test_invalid_split(self) -> None:
         with pytest.raises(AssertionError):

tests/trainers/test_classification.py

@@ -77,6 +77,7 @@ class TestClassificationTask:
             "so2sat_all",
             "so2sat_s1",
             "so2sat_s2",
+            "so2sat_rgb",
             "ucmerced",
         ],
     )