Merge branch 'main' into restart_workers_patch_new

config.yaml

@@ -27,6 +27,23 @@ detect_target:
     model_path: "tests/model_example/yolov8s_ultralytics_pretrained_default.pt"  # See autonomy OneDrive for latest model
     save_prefix: "log_comp"
 
+detect_brightspot:
+    brightspot_percentile_threshold: 99.9
+    filter_by_color: True
+    blob_color: 255
+    filter_by_circularity: False
+    min_circularity: 0.01
+    max_circularity: 1
+    filter_by_inertia: True
+    min_inertia_ratio: 0.2
+    max_inertia_ratio: 1
+    filter_by_convexity: False
+    min_convexity: 0.01
+    max_convexity: 1
+    filter_by_area: True
+    min_area_pixels: 50
+    max_area_pixels: 640
+
 flight_interface:
     # Port 5762 connects directly to the simulated auto pilot, which is more realistic
     # than connecting to port 14550, which is the ground station
@@ -53,6 +70,7 @@ geolocation:
 cluster_estimation:
     min_activation_threshold: 25
     min_new_points_to_run: 5
+    max_num_components: 10
     random_state: 0
 
 communications:

main_2024.py

@@ -137,6 +137,7 @@ def main() -> int:
 
         MIN_ACTIVATION_THRESHOLD = config["cluster_estimation"]["min_activation_threshold"]
         MIN_NEW_POINTS_TO_RUN = config["cluster_estimation"]["min_new_points_to_run"]
+        MAX_NUM_COMPONENTS = config["cluster_estimation"]["max_num_components"]
         RANDOM_STATE = config["cluster_estimation"]["random_state"]
 
         COMMUNICATIONS_TIMEOUT = config["communications"]["timeout"]
@@ -327,7 +328,12 @@ def main() -> int:
     result, cluster_estimation_worker_properties = worker_manager.WorkerProperties.create(
         count=1,
         target=cluster_estimation_worker.cluster_estimation_worker,
-        work_arguments=(MIN_ACTIVATION_THRESHOLD, MIN_NEW_POINTS_TO_RUN, RANDOM_STATE),
+        work_arguments=(
+            MIN_ACTIVATION_THRESHOLD,
+            MIN_NEW_POINTS_TO_RUN,
+            MAX_NUM_COMPONENTS,
+            RANDOM_STATE,
+        ),
         input_queues=[geolocation_to_cluster_estimation_queue],
         output_queues=[cluster_estimation_to_communications_queue],
         controller=controller,

modules/cluster_estimation/cluster_estimation.py

@@ -20,17 +20,6 @@ class ClusterEstimation:
     works by predicting 'cluster centres' from groups of closely placed landing pad
     detections.
 
-    ATTRIBUTES
-    ----------
-    min_activation_threshold: int
-        Minimum total data points before model runs.
-
-    min_new_points_to_run: int
-        Minimum number of new data points that must be collected before running model.
-
-    random_state: int
-        Seed for randomizer, to get consistent results.
-
     METHODS
     -------
     run()
@@ -62,9 +51,6 @@ class ClusterEstimation:
     __MEAN_PRECISION_PRIOR = 1e-6
     __MAX_MODEL_ITERATIONS = 1000
 
-    # Real-world scenario Hyperparameters
-    __MAX_NUM_COMPONENTS = 10  # assumed maximum number of real landing pads
-
     # Hyperparameters to clean up model outputs
     __WEIGHT_DROP_THRESHOLD = 0.1
     __MAX_COVARIANCE_THRESHOLD = 10
@@ -74,24 +60,48 @@ def create(
         cls,
         min_activation_threshold: int,
         min_new_points_to_run: int,
+        max_num_components: int,
         random_state: int,
         local_logger: logger.Logger,
     ) -> "tuple[bool, ClusterEstimation | None]":
         """
         Data requirement conditions for estimation model to run.
+
+        PARAMETERS:
+        min_activation_threshold: int
+            Minimum total data points before model runs. Must be at least max_num_components.
+
+        min_new_points_to_run: int
+            Minimum number of new data points that must be collected before running model. Must be at least 0.
+
+        max_num_components: int
+            Max number of real landing pads. Must be at least 1.
+
+        random_state: int
+            Seed for randomizer, to get consistent results. Must be at least 0.
+
+        local_logger: logger.Logger
+            The local logger to log this object's information.
+
+        RETURNS: The ClusterEstimation object if all conditions pass, otherwise False, None
         """
-        # These parameters must be positive
-        if min_new_points_to_run < 0 or random_state < 0:
+        if min_activation_threshold < max_num_components:
+            return False, None
+
+        if min_new_points_to_run < 0:
+            return False, None
+
+        if max_num_components < 1:
             return False, None
 
-        # At least 1 point for model to fit
-        if min_activation_threshold < 1:
+        if random_state < 0:
             return False, None
 
         return True, ClusterEstimation(
             cls.__create_key,
             min_activation_threshold,
             min_new_points_to_run,
+            max_num_components,
             random_state,
             local_logger,
         )
@@ -101,6 +111,7 @@ def __init__(
         class_private_create_key: object,
         min_activation_threshold: int,
         min_new_points_to_run: int,
+        max_num_components: int,
         random_state: int,
         local_logger: logger.Logger,
     ) -> None:
@@ -112,7 +123,7 @@ def __init__(
         # Initializes VGMM
         self.__vgmm = sklearn.mixture.BayesianGaussianMixture(
             covariance_type=self.__COVAR_TYPE,
-            n_components=self.__MAX_NUM_COMPONENTS,
+            n_components=max_num_components,
             init_params=self.__MODEL_INIT_PARAM,
             weight_concentration_prior=self.__WEIGHT_CONCENTRATION_PRIOR,
             mean_precision_prior=self.__MEAN_PRECISION_PRIOR,

modules/cluster_estimation/cluster_estimation_worker.py

@@ -15,6 +15,7 @@
 def cluster_estimation_worker(
     min_activation_threshold: int,
     min_new_points_to_run: int,
+    max_num_components: int,
     random_state: int,
     input_queue: queue_proxy_wrapper.QueueProxyWrapper,
     output_queue: queue_proxy_wrapper.QueueProxyWrapper,
@@ -31,6 +32,9 @@ def cluster_estimation_worker(
     min_new_points_to_run: int
         Minimum number of new data points that must be collected before running model.
 
+    max_num_components: int
+        Max number of real landing pads.
+
     random_state: int
         Seed for randomizer, to get consistent results.
 
@@ -57,6 +61,7 @@ def cluster_estimation_worker(
     result, estimator = cluster_estimation.ClusterEstimation.create(
         min_activation_threshold,
         min_new_points_to_run,
+        max_num_components,
         random_state,
         local_logger,
     )

modules/detect_target/detect_target_brightspot.py

@@ -13,21 +13,84 @@
 from ..common.modules.logger import logger
 
 
-BRIGHTSPOT_PERCENTILE = 99.9
-
 # Label for brightspots; is 1 since 0 is used for blue landing pads
 DETECTION_LABEL = 1
 # SimpleBlobDetector is a binary detector, so a detection has confidence 1.0 by default
 CONFIDENCE = 1.0
 
 
+# Class has 15 attributes
+# pylint: disable=too-many-instance-attributes
+class DetectTargetBrightspotConfig:
+    """
+    Configuration for DetectTargetBrightspot.
+    """
+
+    def __init__(
+        self,
+        brightspot_percentile_threshold: float,
+        filter_by_color: bool,
+        blob_color: int,
+        filter_by_circularity: bool,
+        min_circularity: float,
+        max_circularity: float,
+        filter_by_inertia: bool,
+        min_inertia_ratio: float,
+        max_inertia_ratio: float,
+        filter_by_convexity: bool,
+        min_convexity: float,
+        max_convexity: float,
+        filter_by_area: bool,
+        min_area_pixels: int,
+        max_area_pixels: int,
+    ) -> None:
+        """
+        Initializes the configuration for DetectTargetBrightspot.
+
+        brightspot_percentile_threshold: Percentile threshold for bright spots.
+        filter_by_color: Whether to filter by color.
+        blob_color: Color of the blob.
+        filter_by_circularity: Whether to filter by circularity.
+        min_circularity: Minimum circularity.
+        max_circularity: Maximum circularity.
+        filter_by_inertia: Whether to filter by inertia.
+        min_inertia_ratio: Minimum inertia ratio.
+        max_inertia_ratio: Maximum inertia ratio.
+        filter_by_convexity: Whether to filter by convexity.
+        min_convexity: Minimum convexity.
+        max_convexity: Maximum convexity.
+        filter_by_area: Whether to filter by area.
+        min_area_pixels: Minimum area in pixels.
+        max_area_pixels: Maximum area in pixels.
+        """
+        self.brightspot_percentile_threshold = brightspot_percentile_threshold
+        self.filter_by_color = filter_by_color
+        self.blob_color = blob_color
+        self.filter_by_circularity = filter_by_circularity
+        self.min_circularity = min_circularity
+        self.max_circularity = max_circularity
+        self.filter_by_inertia = filter_by_inertia
+        self.min_inertia_ratio = min_inertia_ratio
+        self.max_inertia_ratio = max_inertia_ratio
+        self.filter_by_convexity = filter_by_convexity
+        self.min_convexity = min_convexity
+        self.max_convexity = max_convexity
+        self.filter_by_area = filter_by_area
+        self.min_area_pixels = min_area_pixels
+        self.max_area_pixels = max_area_pixels
+
+
+# pylint: enable=too-many-instance-attributes
+
+
 class DetectTargetBrightspot(base_detect_target.BaseDetectTarget):
     """
     Detects bright spots in images.
     """
 
     def __init__(
         self,
+        config: DetectTargetBrightspotConfig,
         local_logger: logger.Logger,
         show_annotations: bool = False,
         save_name: str = "",
@@ -38,6 +101,7 @@ def __init__(
         show_annotations: Display annotated images.
         save_name: Filename prefix for logging detections and annotated images.
         """
+        self.__config = config
         self.__counter = 0
         self.__local_logger = local_logger
         self.__show_annotations = show_annotations
@@ -68,7 +132,9 @@ def run(
             )
             return False, None
 
-        brightspot_threshold = np.percentile(grey_image, BRIGHTSPOT_PERCENTILE)
+        brightspot_threshold = np.percentile(
+            grey_image, self.__config.brightspot_percentile_threshold
+        )
 
         # Apply thresholding to isolate bright spots
         threshold_used, bw_image = cv2.threshold(
@@ -80,14 +146,20 @@ def run(
 
         # Set up SimpleBlobDetector
         params = cv2.SimpleBlobDetector_Params()
-        params.filterByColor = True
-        params.blobColor = 255
-        params.filterByCircularity = False
-        params.filterByInertia = True
-        params.minInertiaRatio = 0.2
-        params.filterByConvexity = False
-        params.filterByArea = True
-        params.minArea = 50  # pixels
+        params.filterByColor = self.__config.filter_by_color
+        params.blobColor = self.__config.blob_color
+        params.filterByCircularity = self.__config.filter_by_circularity
+        params.minCircularity = self.__config.min_circularity
+        params.maxCircularity = self.__config.max_circularity
+        params.filterByInertia = self.__config.filter_by_inertia
+        params.minInertiaRatio = self.__config.min_inertia_ratio
+        params.maxInertiaRatio = self.__config.max_inertia_ratio
+        params.filterByConvexity = self.__config.filter_by_convexity
+        params.minConvexity = self.__config.min_convexity
+        params.maxConvexity = self.__config.max_convexity
+        params.filterByArea = self.__config.filter_by_area
+        params.minArea = self.__config.min_area_pixels
+        params.maxArea = self.__config.max_area_pixels
 
         detector = cv2.SimpleBlobDetector_create(params)
         keypoints = detector.detect(bw_image)

tests/brightspot_example/generate_expected.py

@@ -32,6 +32,24 @@
     pathlib.Path(f"ir_no_detections_{i}.png") for i in range(0, NUMBER_OF_IMAGES_NO_DETECTIONS)
 ]
 
+DETECT_TARGET_BRIGHTSPOT_CONFIG = detect_target_brightspot.DetectTargetBrightspotConfig(
+    brightspot_percentile_threshold=99.9,
+    filter_by_color=True,
+    blob_color=255,
+    filter_by_circularity=False,
+    min_circularity=0.01,
+    max_circularity=1,
+    filter_by_inertia=True,
+    min_inertia_ratio=0.2,
+    max_inertia_ratio=1,
+    filter_by_convexity=False,
+    min_convexity=0.01,
+    max_convexity=1,
+    filter_by_area=True,
+    min_area_pixels=50,
+    max_area_pixels=640,
+)
+
 
 def main() -> int:
     """
@@ -43,7 +61,10 @@ def main() -> int:
         return 1
 
     detector = detect_target_brightspot.DetectTargetBrightspot(
-        local_logger=temp_logger, show_annotations=False, save_name=""
+        config=DETECT_TARGET_BRIGHTSPOT_CONFIG,
+        local_logger=temp_logger,
+        show_annotations=False,
+        save_name="",
     )
 
     for image_file, annotated_image_path, expected_detections_path in zip(

tests/unit/test_cluster_detection.py

@@ -13,10 +13,10 @@
 
 MIN_TOTAL_POINTS_THRESHOLD = 100
 MIN_NEW_POINTS_TO_RUN = 10
+MAX_NUM_COMPONENTS = 10
 RNG_SEED = 0
 CENTRE_BOX_SIZE = 500
 
-
 # Test functions use test fixture signature names and access class privates
 # No enable
 # pylint: disable=protected-access,redefined-outer-name
@@ -34,6 +34,7 @@ def cluster_model() -> cluster_estimation.ClusterEstimation:  # type: ignore
     result, model = cluster_estimation.ClusterEstimation.create(
         MIN_TOTAL_POINTS_THRESHOLD,
         MIN_NEW_POINTS_TO_RUN,
+        MAX_NUM_COMPONENTS,
         RNG_SEED,
         test_logger,
     )