Merge branch 'main' into duckdb

.github/workflows/test.yml

@@ -385,6 +385,7 @@ jobs:
       - name: Upload to Codecov
         uses: codecov/codecov-action@v3
         with:
+          token: ${{ secrets.CODECOV_TOKEN }}
           verbose: true
 
       # - name: Download Coverage

Dockerfile

@@ -9,13 +9,6 @@ RUN apt-get  update --yes --quiet
 RUN apt-get install --yes --quiet --no-install-recommends \
     curl ffmpeg git
 
-RUN sed -i 's/#force_color_prompt/force_color_prompt/g' $HOME/.bashrc
-RUN git clone --recurse-submodules --depth 1 https://github.com/chanwutk/.config.git $HOME/.config
-RUN echo "source $HOME/.config/rc" >> $HOME/.bashrc
-RUN cat <(echo "source $HOME/.config/profile") $HOME/.profile > $HOME/.profile.tmp
-RUN mv $HOME/.profile.tmp $HOME/.profile
-
-
 # configre python to output directly to terminal
 # see: https://stackoverflow.com/questions/59812009/what-is-the-use-of-pythonunbuffered-in-docker-file
 ENV PYTHONUNBUFFERED=1

README.md

@@ -34,7 +34,7 @@
   >
 </p>
 
-## Absract
+## Abstract
 Videos that are shot using commodity hardware such as phones and surveillance cameras record various metadata such as time and location.
 We encounter such geospatial videos on a daily basis and such videos have been growing in volume significantly.
 Yet, we do not have data management systems that allow users to interact with such data effectively.
@@ -122,7 +122,7 @@ jupyter-lab
 The demo notebook first constructs the world. Then it queries for the trajectory of the cars that appeared once in an area of interests within some time interval.
 
 ## Citing Spatialyze
-This paper is currently under review for [VLDB](https://vldb.org/2024/).
+This paper will be presented at [VLDB](https://vldb.org/2024/).
 ```bib
 @misc{kittivorawong2023spatialyze,
     title={Spatialyze: A Geospatial Video Analytics System with Spatial-Aware Optimizations}, 

pyproject.toml

@@ -23,7 +23,7 @@ torch = "^1.13.0"
 torchvision = "^0.14.0"
 matplotlib = "^3.5.2"
 scipy = "^1.4.1"
-gdown = "^4.7.3"
+gdown = "^5.1.0"
 ultralytics = "^8.0.148"
 notebook = "^7.0.6"
 

spatialyze/database.py

@@ -32,8 +32,8 @@
 from .video_processor.types import Float33
 
 if TYPE_CHECKING:
-    from psycopg2 import connection as Connection
-    from psycopg2 import cursor as Cursor
+    from psycopg2._psycopg import connection as Connection
+    from psycopg2._psycopg import cursor as Cursor
 
     from .predicate import PredicateNode
 
@@ -313,6 +313,7 @@ def predicate(self, predicate: "PredicateNode", temporal: bool = True):
     def sql(self, query: str) -> pd.DataFrame:
         results, cursor = self.execute_and_cursor(query)
         description = cursor.description
+        assert description is not None
         cursor.close()
         return pd.DataFrame(results, columns=[d.name for d in description])
 

spatialyze/predicate.py

@@ -91,11 +91,11 @@ def __or__(self, other):
             ],
         )
 
-    def __eq__(self, other):
+    def __eq__(self, other):  # pyright: ignore [reportIncompatibleMethodOverride]
         other = wrap_literal(other)
         return CompOpNode(self, "eq", other)
 
-    def __ne__(self, other):
+    def __ne__(self, other):  # pyright: ignore [reportIncompatibleMethodOverride]
         other = wrap_literal(other)
         return CompOpNode(self, "ne", other)
 

spatialyze/video_processor/stages/decode_frame/parallel_decode_frame.py

@@ -63,4 +63,6 @@ def _r(acc: "tuple[int, list[tuple[int, int]]]", frames: int):
             return None, {self.classname(): metadata}
         except BaseException:
             _, output = DecodeFrame()._run(payload)
-            return None, {self.classname(): DecodeFrame.get(output)}
+            images = DecodeFrame.get(output)
+            assert images is not None
+            return None, {self.classname(): images}

spatialyze/video_processor/stages/depth_estimation.py

@@ -113,7 +113,9 @@ def eval_all(self, input_images: "list[npt.NDArray | None]"):
                 # Load image and preprocess
                 input_image = pil.fromarray(im[:, :, [2, 1, 0]])
                 original_width, original_height = input_image.size
-                input_image = input_image.resize((self.feed_width, self.feed_height), pil.LANCZOS)
+                input_image = input_image.resize(
+                    (self.feed_width, self.feed_height), pil.Resampling.LANCZOS
+                )
                 input_image = transforms.ToTensor()(input_image).unsqueeze(0)
 
                 # PREDICTION

spatialyze/video_processor/stages/detection_2d/ground_truth.py

@@ -189,7 +189,7 @@ def __init__(self, df_annotations: "pd.DataFrame"):
         self.class_to_id = {c: i for i, c in enumerate(self.id_to_classes)}
 
     def _run(self, payload: "Payload"):
-        metadata: "list[Metadatum | None]" = []
+        metadata: list[Metadatum] = []
         dimension = payload.video.dimension
         for i, cc in enumerate(payload.video._camera_configs):
             fid = cc.frame_id
@@ -224,7 +224,9 @@ def _run(self, payload: "Payload"):
             else:
                 metadata.append(
                     Metadatum(
-                        torch.Tensor(tensor), yolo_classes, [DetectionId(i, _id) for _id in ids]
+                        torch.Tensor(tensor),
+                        yolo_classes,
+                        [DetectionId(i, _id) for _id in ids],
                     )
                 )
 

spatialyze/video_processor/stages/detection_3d/ground_truth.py

@@ -27,7 +27,7 @@ def __init__(self, df_annotations: "pd.DataFrame"):
         self.class_to_id = {c: i for i, c in enumerate(self.id_to_classes)}
 
     def _run(self, payload: "Payload"):
-        metadata: "list[Metadatum | None]" = []
+        metadata: list[Metadatum] = []
         dimension = payload.video.dimension
         for i, cc in enumerate(payload.video._camera_configs):
             fid = cc.frame_id
@@ -81,7 +81,9 @@ def _run(self, payload: "Payload"):
             else:
                 metadata.append(
                     Metadatum(
-                        torch.Tensor(tensor), yolo_classes, [DetectionId(i, _id) for _id in ids]
+                        torch.Tensor(tensor),
+                        yolo_classes,
+                        [DetectionId(i, _id) for _id in ids],
                     )
                 )
 

spatialyze/video_processor/stages/stage.py

@@ -90,4 +90,6 @@ def tqdm(cls, iterable: "Iterable[_T2]", *args, **kwargs) -> "Iterable[_T2]":
 def _get_classnames(cls: "type") -> "list[str]":
     if cls == Stage:
         return []
-    return [*_get_classnames(cls.__base__), cls.__name__]
+    base = cls.__base__
+    assert base is not None
+    return [*_get_classnames(base), cls.__name__]

spatialyze/video_processor/stream/mono_depth_estimator.py

@@ -25,7 +25,9 @@ def _stream(self, video: Video):
                 # Load image and preprocess
                 input_image = Image.fromarray(img[:, :, [2, 1, 0]])
                 original_width, original_height = input_image.size
-                input_image = input_image.resize((md.feed_width, md.feed_height), Image.LANCZOS)
+                input_image = input_image.resize(
+                    (md.feed_width, md.feed_height), Image.Resampling.LANCZOS
+                )
                 input_image = transforms.ToTensor()(input_image).unsqueeze(0)
 
                 # PREDICTION