Fix typing for Python 3.8 (#22)

* Fix typing for Python 3.8

* Fix set typing

* Add changelog and bump version

.github/workflows/ci.yaml

@@ -31,7 +31,7 @@ jobs:
     strategy:
       matrix:
         operating-system: [ubuntu-latest, macos-latest]
-        python-version: ["3.9", "3.10"]
+        python-version: ["3.8", "3.9", "3.10"]
       fail-fast: false
     steps:
       - name: Checkout

CHANGELOG.md

@@ -4,6 +4,13 @@ All notable changes to this project will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
+
+## [0.1.7] - 2022-10-02
+
+## Fixed
+
+* Fix typing for compatibility with Python 3.8 ([#22](https://github.com/tadejsv/EvalDeT/pull/22))
+
 ## [0.1.6] - 2022-09-29
 
 ## Changed

pyproject.toml

@@ -14,6 +14,7 @@ keywords = ["evaluation", "tracking", "object detection", "computer vision"]
 classifiers = [
     "License :: OSI Approved :: Apache Software License",
     "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.8",
     "Programming Language :: Python :: 3.9",
     "Programming Language :: Python :: 3.10",
     "Development Status :: 4 - Beta",

src/evaldet/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.1.6"
+__version__ = "0.1.7"
 
 from .metrics import MOTMetrics  # noqa: F401
 from .tracks import Tracks  # noqa: F401

src/evaldet/metrics.py

@@ -29,8 +29,8 @@ class MOTMetrics(CLEARMOTMetrics, IDMetrics, HOTAMetrics):
     efficiently share pre-computed IoU distances.
     """
 
-    _ious: list[np.ndarray]
-    _ious_dict: dict[int, int]
+    _ious: t.List[np.ndarray]
+    _ious_dict: t.Dict[int, int]
 
     def __init__(
         self, clearmot_dist_threshold: float = 0.5, id_dist_threshold: float = 0.5

src/evaldet/mot_metrics/clearmot.py

@@ -31,11 +31,11 @@ def _calculate_clearmot_metrics(
         ground_truths = 0
 
         matches_dist = []
-        matching: dict[int, int] = {}
+        matching: t.Dict[int, int] = {}
 
         # This is the persistent matching dictionary, used to check for mismatches
         # when a previously matched hypothesis is re-matched with a ground truth
-        matching_persist: dict[int, int] = {}
+        matching_persist: t.Dict[int, int] = {}
 
         for frame in all_frames:
             if frame not in ground_truth:

src/evaldet/mot_metrics/hota.py

@@ -24,7 +24,7 @@ class HOTAResults(t.TypedDict):
 
 
 def _create_coo_array(
-    vals_list: dict[tuple[int, int], int], shape: tuple[int, int]
+    vals_list: t.Dict[t.Tuple[int, int], int], shape: t.Tuple[int, int]
 ) -> sparse.coo_array:
     row_inds = np.array(tuple(x[0] for x in vals_list.keys()))
     col_inds = np.array(tuple(x[1] for x in vals_list.keys()))
@@ -54,7 +54,7 @@ def _calculate_hota_metrics(
         TPA_max = np.zeros((len(alphas), n_gt, n_hyp), dtype=np.int32)
         FPA_max = np.tile(np.tile(hyps_counts, (n_gt, 1)), (len(alphas), 1, 1))
         FNA_max = np.tile(np.tile(gts_counts, (n_hyp, 1)).T, (len(alphas), 1, 1))
-        TPA_max_vals: list[dict[tuple[int, int], int]] = [
+        TPA_max_vals: t.List[t.Dict[t.Tuple[int, int], int]] = [
             co.defaultdict(int) for _ in range(len(alphas))
         ]
 
@@ -85,7 +85,7 @@ def _calculate_hota_metrics(
         A_max = TPA_max / (FNA_max + FPA_max - TPA_max)
 
         # Do the actual matching
-        TPA_vals: list[dict[tuple[int, int], int]] = [
+        TPA_vals: t.List[t.Dict[t.Tuple[int, int], int]] = [
             co.defaultdict(int) for _ in range(len(alphas))
         ]
         for frame in sorted(set(ground_truth.frames).intersection(hypotheses.frames)):

src/evaldet/tracks.py

@@ -1,4 +1,5 @@
 import csv
+import typing as t
 import xml.etree.ElementTree as ET
 from pathlib import Path
 from typing import Any, NamedTuple, Optional, Union
@@ -43,7 +44,7 @@ class Tracks:
     def from_csv(
         cls,
         csv_file: Union[str, Path],
-        fieldnames: list[str],
+        fieldnames: t.List[str],
         zero_indexed: bool = True,
     ) -> "Tracks":
         """Get detections from a CSV file.
@@ -72,7 +73,7 @@ def from_csv(
         tracks = cls()
         with open(csv_file, newline="") as file:
             csv_reader = csv.DictReader(file, fieldnames=fieldnames, dialect="unix")
-            frames: dict[int, Any] = {}
+            frames: t.Dict[int, Any] = {}
 
             for line_num, line in enumerate(csv_reader):
                 try:
@@ -212,7 +213,7 @@ def from_ua_detrac(
         cls,
         file_path: Union[Path, str],
         classes_attr_name: Optional[str] = None,
-        classes_list: Optional[list[str]] = None,
+        classes_list: Optional[t.List[str]] = None,
     ) -> "Tracks":
         """Creates a Tracks object from detections file in the UA-DETRAC XML format.
 
@@ -282,9 +283,9 @@ def from_ua_detrac(
             tracks_f = frame.find("target_list").findall("target")  # type: ignore
 
             current_frame = int(frame.attrib["num"])
-            detections: list[list[float]] = []
-            classes: list[int] = []
-            ids: list[int] = []
+            detections: t.List[t.List[float]] = []
+            classes: t.List[int] = []
+            ids: t.List[int] = []
 
             for track in tracks_f:
                 # Get track attributes
@@ -315,7 +316,7 @@ def from_ua_detrac(
     def from_cvat_video(
         cls,
         file_path: Union[Path, str],
-        classes_list: list[str],
+        classes_list: t.List[str],
     ) -> "Tracks":
         """Creates a Tracks object from detections file in the CVAT for Video XML
         format.
@@ -366,7 +367,7 @@ def from_cvat_video(
         root = xml_tree.getroot()
         tracks = cls()
 
-        frames: dict[int, Any] = {}
+        frames: t.Dict[int, Any] = {}
         tracks_cvat = root.findall("track")
         for track_cvat in tracks_cvat:
             track_id = int(track_cvat.attrib[_ID_KEY])
@@ -438,19 +439,19 @@ def _add_to_tracks_accumulator(frames: dict, new_obj: dict) -> None:
 
     def __init__(self) -> None:
 
-        self._frame_nums: set[int] = set()
-        self._detections: dict[int, np.ndarray] = dict()
-        self._ids: dict[int, np.ndarray] = dict()
-        self._classes: dict[int, np.ndarray] = dict()
-        self._confs: dict[int, np.ndarray] = dict()
+        self._frame_nums: t.Set[int] = set()
+        self._detections: t.Dict[int, np.ndarray] = dict()
+        self._ids: t.Dict[int, np.ndarray] = dict()
+        self._classes: t.Dict[int, np.ndarray] = dict()
+        self._confs: t.Dict[int, np.ndarray] = dict()
 
     def add_frame(
         self,
         frame_num: int,
-        ids: Union[list[int], np.ndarray],
+        ids: Union[t.List[int], np.ndarray],
         detections: np.ndarray,
-        classes: Optional[Union[list[int], np.ndarray]] = None,
-        confs: Optional[Union[list[float], np.ndarray]] = None,
+        classes: Optional[Union[t.List[int], np.ndarray]] = None,
+        confs: Optional[Union[t.List[float], np.ndarray]] = None,
     ) -> None:
         """Add a frame to the collection. Can overwrite existing frame.
 
@@ -562,7 +563,7 @@ def filter_frame(self, frame_num: int, filter: np.ndarray) -> None:
         if frame.classes is not None:
             self._classes[frame_num] = frame.classes[filter]
 
-    def filter_by_class(self, classes: list[int]) -> None:
+    def filter_by_class(self, classes: t.List[int]) -> None:
         """Filter all frames by classes
 
         This will keep the detections with class label corresponding to one of the
@@ -604,32 +605,32 @@ def filter_by_conf(self, lower_bound: float) -> None:
                 self.filter_frame(frame, filter_conf)
 
     @property
-    def all_classes(self) -> set[int]:
+    def all_classes(self) -> t.Set[int]:
         """Get a set of all classes in the collection."""
-        classes: set[int] = set()
+        classes: t.Set[int] = set()
         for frame in self._frame_nums:
             if frame in self._classes:
                 classes.update(self._classes[frame])
 
         return classes
 
     @property
-    def ids_count(self) -> dict[int, int]:
+    def ids_count(self) -> t.Dict[int, int]:
         """Get the number of frames that each id is present in.
 
         Returns:
             A dictionary where keys are the track ids, and values
             are the numbers of frames they appear in.
         """
-        ids_count: dict[int, int] = dict()
+        ids_count: t.Dict[int, int] = dict()
         for frame in self._frame_nums:
             for _id in self._ids[frame]:
                 ids_count[_id] = ids_count.get(_id, 0) + 1
 
         return ids_count
 
     @property
-    def frames(self) -> set[int]:
+    def frames(self) -> t.Set[int]:
         """Get an ordered list of all frame numbers in the collection."""
         return self._frame_nums.copy()