Merge branch 'fix_detectron2' of github.com:opendr-eu/opendr into fix…

…_detectron2

tests/sources/tools/perception/object_tracking_2d/fair_mot/test_object_tracking_2d_fair_mot.py

@@ -84,66 +84,66 @@ def tearDownClass(cls):
 
         rmdir(os.path.join(cls.temp_dir))
 
-#     def test_fit(self):
-
-#         def test_model(name):
-#             dataset = MotDataset(self.dataset_path)
-
-#             learner = ObjectTracking2DFairMotLearner(
-#                 iters=1,
-#                 num_epochs=1,
-#                 checkpoint_after_iter=3,
-#                 temp_path=self.temp_dir,
-#                 device=DEVICE,
-#             )
-
-#             starting_param = list(learner.model.parameters())[0].clone()
-
-#             learner.fit(
-#                 dataset,
-#                 val_epochs=-1,
-#                 train_split_paths=self.train_split_paths,
-#                 val_split_paths=self.train_split_paths,
-#                 verbose=True,
-#             )
-#             new_param = list(learner.model.parameters())[0].clone()
-#             self.assertFalse(torch.equal(starting_param, new_param))
-
-#             print("Fit", name, "ok", file=sys.stderr)
-
-#         for name in self.model_names:
-#             test_model(name)
-
-#     def test_fit_iterator(self):
-#         def test_model(name):
-#             dataset = MotDatasetIterator(self.dataset_path, self.train_split_paths)
-#             eval_dataset = RawMotDatasetIterator(self.dataset_path, self.train_split_paths)
-
-#             learner = ObjectTracking2DFairMotLearner(
-#                 iters=1,
-#                 num_epochs=1,
-#                 checkpoint_after_iter=3,
-#                 temp_path=self.temp_dir,
-#                 device=DEVICE,
-#             )
-
-#             starting_param = list(learner.model.parameters())[0].clone()
-
-#             learner.fit(
-#                 dataset,
-#                 val_dataset=eval_dataset,
-#                 val_epochs=-1,
-#                 train_split_paths=self.train_split_paths,
-#                 val_split_paths=self.train_split_paths,
-#                 verbose=True,
-#             )
-#             new_param = list(learner.model.parameters())[0].clone()
-#             self.assertFalse(torch.equal(starting_param, new_param))
-
-#             print("Fit iterator", name, "ok", file=sys.stderr)
-
-#         for name in self.model_names:
-#             test_model(name)
+    def test_fit(self):
+
+        def test_model(name):
+            dataset = MotDataset(self.dataset_path)
+
+            learner = ObjectTracking2DFairMotLearner(
+                iters=1,
+                num_epochs=1,
+                checkpoint_after_iter=3,
+                temp_path=self.temp_dir,
+                device=DEVICE,
+            )
+
+            starting_param = list(learner.model.parameters())[0].clone()
+
+            learner.fit(
+                dataset,
+                val_epochs=-1,
+                train_split_paths=self.train_split_paths,
+                val_split_paths=self.train_split_paths,
+                verbose=True,
+            )
+            new_param = list(learner.model.parameters())[0].clone()
+            self.assertFalse(torch.equal(starting_param, new_param))
+
+            print("Fit", name, "ok", file=sys.stderr)
+
+        for name in self.model_names:
+            test_model(name)
+
+    def test_fit_iterator(self):
+        def test_model(name):
+            dataset = MotDatasetIterator(self.dataset_path, self.train_split_paths)
+            eval_dataset = RawMotDatasetIterator(self.dataset_path, self.train_split_paths)
+
+            learner = ObjectTracking2DFairMotLearner(
+                iters=1,
+                num_epochs=1,
+                checkpoint_after_iter=3,
+                temp_path=self.temp_dir,
+                device=DEVICE,
+            )
+
+            starting_param = list(learner.model.parameters())[0].clone()
+
+            learner.fit(
+                dataset,
+                val_dataset=eval_dataset,
+                val_epochs=-1,
+                train_split_paths=self.train_split_paths,
+                val_split_paths=self.train_split_paths,
+                verbose=True,
+            )
+            new_param = list(learner.model.parameters())[0].clone()
+            self.assertFalse(torch.equal(starting_param, new_param))
+
+            print("Fit iterator", name, "ok", file=sys.stderr)
+
+        for name in self.model_names:
+            test_model(name)
 
     def test_eval(self):
         def test_model(name):