ObjectDetectionTask: increase test coverage for torchvision 0.14+

tests/trainers/test_classification.py

@@ -61,10 +61,14 @@ def load(url: str, *args: Any, **kwargs: Any) -> dict[str, Any]:
     return state_dict
 
 
-def plot(*args: Any, **kwargs: Any) -> None:
+def plot_no_rgb(*args: Any, **kwargs: Any) -> None:
     raise ValueError
 
 
+def no_plot_method(*args: Any, **kwargs: Any) -> None:
+    return None
+
+
 class TestClassificationTask:
     @pytest.mark.parametrize(
         "name",
@@ -181,7 +185,21 @@ def test_invalid_loss(self) -> None:
             ClassificationTask(model="resnet18", loss="invalid_loss")
 
     def test_no_rgb(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
-        monkeypatch.setattr(EuroSATDataModule, "plot", plot)
+        monkeypatch.setattr(EuroSATDataModule, "plot", plot_no_rgb)
+        datamodule = EuroSATDataModule(
+            root="tests/data/eurosat", batch_size=1, num_workers=0
+        )
+        model = ClassificationTask(model="resnet18", in_channels=13, num_classes=10)
+        trainer = Trainer(
+            accelerator="cpu",
+            fast_dev_run=fast_dev_run,
+            log_every_n_steps=1,
+            max_epochs=1,
+        )
+        trainer.validate(model=model, datamodule=datamodule)
+
+    def test_no_plot_method(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
+        monkeypatch.setattr(EuroSATDataModule, "plot", no_plot_method)
         datamodule = EuroSATDataModule(
             root="tests/data/eurosat", batch_size=1, num_workers=0
         )
@@ -258,7 +276,23 @@ def test_invalid_loss(self) -> None:
             MultiLabelClassificationTask(model="resnet18", loss="invalid_loss")
 
     def test_no_rgb(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
-        monkeypatch.setattr(BigEarthNetDataModule, "plot", plot)
+        monkeypatch.setattr(BigEarthNetDataModule, "plot", plot_no_rgb)
+        datamodule = BigEarthNetDataModule(
+            root="tests/data/bigearthnet", batch_size=1, num_workers=0
+        )
+        model = MultiLabelClassificationTask(
+            model="resnet18", in_channels=14, num_classes=19, loss="bce"
+        )
+        trainer = Trainer(
+            accelerator="cpu",
+            fast_dev_run=fast_dev_run,
+            log_every_n_steps=1,
+            max_epochs=1,
+        )
+        trainer.validate(model=model, datamodule=datamodule)
+
+    def test_no_plot_method(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
+        monkeypatch.setattr(BigEarthNetDataModule, "plot", no_plot_method)
         datamodule = BigEarthNetDataModule(
             root="tests/data/bigearthnet", batch_size=1, num_workers=0
         )

tests/trainers/test_detection.py

@@ -45,20 +45,27 @@ def forward(self, images: Any, targets: Any = None) -> Any:
         else:  # eval mode
             output = []
             for i in range(batch_size):
+                boxes = torch.zeros(10, 4, dtype=torch.float)
+                # Create xmax, ymax larger than 0.0
+                boxes[:, 2:4] = torch.FloatTensor(10, 2).uniform_(0.1, 0.9)
                 output.append(
                     {
-                        "boxes": torch.rand(10, 4),
+                        "boxes": boxes,
                         "labels": torch.randint(0, 2, (10,)),
                         "scores": torch.rand(10),
                     }
                 )
             return output
 
 
-def plot(*args: Any, **kwargs: Any) -> None:
+def plot_no_rgb(*args: Any, **kwargs: Any) -> None:
     raise ValueError
 
 
+def no_plot_method(*args: Any, **kwargs: Any) -> None:
+    return None
+
+
 class TestObjectDetectionTask:
     @pytest.mark.parametrize("name", ["nasa_marine_debris"])
     @pytest.mark.parametrize("model_name", ["faster-rcnn", "fcos", "retinanet"])
@@ -114,7 +121,21 @@ def test_pretrained_backbone(self) -> None:
         ObjectDetectionTask(backbone="resnet18", weights=True)
 
     def test_no_rgb(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
-        monkeypatch.setattr(NASAMarineDebrisDataModule, "plot", plot)
+        monkeypatch.setattr(NASAMarineDebrisDataModule, "plot", plot_no_rgb)
+        datamodule = NASAMarineDebrisDataModule(
+            root="tests/data/nasa_marine_debris", batch_size=1, num_workers=0
+        )
+        model = ObjectDetectionTask(backbone="resnet18", num_classes=2)
+        trainer = Trainer(
+            accelerator="cpu",
+            fast_dev_run=fast_dev_run,
+            log_every_n_steps=1,
+            max_epochs=1,
+        )
+        trainer.validate(model=model, datamodule=datamodule)
+
+    def test_no_plot_method(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
+        monkeypatch.setattr(NASAMarineDebrisDataModule, "plot", no_plot_method)
         datamodule = NASAMarineDebrisDataModule(
             root="tests/data/nasa_marine_debris", batch_size=1, num_workers=0
         )

tests/trainers/test_regression.py

@@ -51,10 +51,14 @@ def load(url: str, *args: Any, **kwargs: Any) -> dict[str, Any]:
     return state_dict
 
 
-def plot(*args: Any, **kwargs: Any) -> None:
+def plot_no_rgb(*args: Any, **kwargs: Any) -> None:
     raise ValueError
 
 
+def no_plot_method(*args: Any, **kwargs: Any) -> None:
+    return None
+
+
 class TestRegressionTask:
     @classmethod
     def create_model(*args: Any, **kwargs: Any) -> Module:
@@ -153,7 +157,21 @@ def test_weight_str_download(self, weights: WeightsEnum) -> None:
         )
 
     def test_no_rgb(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
-        monkeypatch.setattr(TropicalCycloneDataModule, "plot", plot)
+        monkeypatch.setattr(TropicalCycloneDataModule, "plot", plot_no_rgb)
+        datamodule = TropicalCycloneDataModule(
+            root="tests/data/cyclone", batch_size=1, num_workers=0
+        )
+        model = RegressionTask(model="resnet18")
+        trainer = Trainer(
+            accelerator="cpu",
+            fast_dev_run=fast_dev_run,
+            log_every_n_steps=1,
+            max_epochs=1,
+        )
+        trainer.validate(model=model, datamodule=datamodule)
+
+    def test_no_plot_method(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
+        monkeypatch.setattr(TropicalCycloneDataModule, "plot", no_plot_method)
         datamodule = TropicalCycloneDataModule(
             root="tests/data/cyclone", batch_size=1, num_workers=0
         )

tests/trainers/test_segmentation.py

@@ -43,10 +43,14 @@ def load(url: str, *args: Any, **kwargs: Any) -> dict[str, Any]:
     return state_dict
 
 
-def plot(*args: Any, **kwargs: Any) -> None:
+def plot_no_rgb(*args: Any, **kwargs: Any) -> None:
     raise ValueError
 
 
+def no_plot_method(*args: Any, **kwargs: Any) -> None:
+    return None
+
+
 class TestSemanticSegmentationTask:
     @pytest.mark.parametrize(
         "name",
@@ -180,7 +184,23 @@ def test_ignoreindex_with_jaccard(self) -> None:
             SemanticSegmentationTask(loss="jaccard", ignore_index=0)
 
     def test_no_rgb(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
-        monkeypatch.setattr(SEN12MSDataModule, "plot", plot)
+        monkeypatch.setattr(SEN12MSDataModule, "plot", plot_no_rgb)
+        datamodule = SEN12MSDataModule(
+            root="tests/data/sen12ms", batch_size=1, num_workers=0
+        )
+        model = SemanticSegmentationTask(
+            backbone="resnet18", in_channels=15, num_classes=6
+        )
+        trainer = Trainer(
+            accelerator="cpu",
+            fast_dev_run=fast_dev_run,
+            log_every_n_steps=1,
+            max_epochs=1,
+        )
+        trainer.validate(model=model, datamodule=datamodule)
+
+    def test_no_plot_method(self, monkeypatch: MonkeyPatch, fast_dev_run: bool) -> None:
+        monkeypatch.setattr(SEN12MSDataModule, "plot", no_plot_method)
         datamodule = SEN12MSDataModule(
             root="tests/data/sen12ms", batch_size=1, num_workers=0
         )