Missing types and pyleft (#160)

.github/workflows/ci.yaml

@@ -64,7 +64,7 @@ jobs:
     - name: Install dependencies
       run: pip install -e .[dev]
 
-    - name: Type check
+    - name: Pyright check
       uses: jakebailey/pyright-action@v2
       with:
         version: '1.1.380'
@@ -73,6 +73,9 @@ jobs:
         python-version: '3.10'
         project: pyproject.toml
 
+    - name: Pyleft Check
+      run: pyleft luxonis_train
+
   tests:
     needs:
       - type-check

luxonis_train/attached_modules/losses/adaptive_detection_loss.py

@@ -1,4 +1,4 @@
-from typing import Any, Literal, cast
+from typing import Literal, cast
 
 import torch
 import torch.nn.functional as F
@@ -37,7 +37,7 @@ def __init__(
         reduction: Literal["sum", "mean"] = "mean",
         class_loss_weight: float = 1.0,
         iou_loss_weight: float = 2.5,
-        **kwargs: Any,
+        **kwargs,
     ):
         """BBox loss adapted from U{YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications
         <https://arxiv.org/pdf/2209.02976.pdf>}. It combines IoU based bbox regression loss and varifocal loss

luxonis_train/attached_modules/losses/bce_with_logits.py

@@ -1,4 +1,4 @@
-from typing import Any, Literal
+from typing import Literal
 
 import torch
 from torch import Tensor, nn
@@ -16,7 +16,7 @@ def __init__(
         weight: list[float] | None = None,
         reduction: Literal["none", "mean", "sum"] = "mean",
         pos_weight: Tensor | None = None,
-        **kwargs: Any,
+        **kwargs,
     ):
         """This loss combines a L{nn.Sigmoid} layer and the
         L{nn.BCELoss} in one single class. This version is more

luxonis_train/attached_modules/losses/cross_entropy.py

@@ -1,4 +1,4 @@
-from typing import Any, Literal
+from typing import Literal
 
 import torch
 import torch.nn as nn
@@ -22,7 +22,7 @@ def __init__(
         ignore_index: int = -100,
         reduction: Literal["none", "mean", "sum"] = "mean",
         label_smoothing: float = 0.0,
-        **kwargs: Any,
+        **kwargs,
     ):
         super().__init__(**kwargs)
 

luxonis_train/attached_modules/losses/ctc_loss.py

@@ -1,5 +1,3 @@
-from typing import Any
-
 import torch
 import torch.nn as nn
 from torch import Tensor
@@ -14,7 +12,7 @@ class CTCLoss(BaseLoss):
 
     node: OCRCTCHead
 
-    def __init__(self, use_focal_loss: bool = True, **kwargs: Any):
+    def __init__(self, use_focal_loss: bool = True, **kwargs):
         """Initializes the CTC loss with optional focal loss support.
 
         @type use_focal_loss: bool

luxonis_train/attached_modules/losses/efficient_keypoint_bbox_loss.py

@@ -1,4 +1,4 @@
-from typing import Any, Literal
+from typing import Literal
 
 import torch
 import torch.nn.functional as F
@@ -37,7 +37,7 @@ def __init__(
         vis_kpts_loss_weight: float = 1.0,
         sigmas: list[float] | None = None,
         area_factor: float | None = None,
-        **kwargs: Any,
+        **kwargs,
     ):
         """BBox loss adapted from U{YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications
         <https://arxiv.org/pdf/2209.02976.pdf>}. It combines IoU based bbox regression loss and varifocal loss

luxonis_train/attached_modules/losses/fomo_localization_loss.py

@@ -1,5 +1,3 @@
-from typing import Any
-
 import torch
 import torch.nn.functional as F
 from torch import Tensor
@@ -15,13 +13,11 @@ class FOMOLocalizationLoss(BaseLoss):
     node: FOMOHead
     supported_tasks = [Tasks.FOMO]
 
-    def __init__(self, object_weight: float = 500, **kwargs: Any):
+    def __init__(self, object_weight: float = 500, **kwargs):
         """FOMO Localization Loss for object detection using heatmaps.
 
         @type object_weight: float
         @param object_weight: Weight for object in loss calculation.
-        @type kwargs: Any
-        @param kwargs: Additional arguments.
         """
         super().__init__(**kwargs)
         self.original_img_size = self.original_in_shape[1:]

luxonis_train/attached_modules/losses/ohem_loss.py

@@ -1,5 +1,3 @@
-from typing import Any
-
 import torch
 from torch import Tensor
 
@@ -18,7 +16,7 @@ def __init__(
         criterion: BaseLoss,
         ohem_ratio: float = 0.1,
         ohem_threshold: float = 0.7,
-        **kwargs: Any,
+        **kwargs,
     ):
         """Initializes the criterion.
 

luxonis_train/attached_modules/losses/precision_dfl_detection_loss.py

@@ -1,4 +1,4 @@
-from typing import Any, cast
+from typing import cast
 
 import torch
 import torch.nn.functional as F
@@ -28,7 +28,7 @@ def __init__(
         class_loss_weight: float = 0.5,
         bbox_loss_weight: float = 7.5,
         dfl_loss_weight: float = 1.5,
-        **kwargs: Any,
+        **kwargs,
     ):
         """BBox loss adapted from  U{Real-Time Flying Object Detection with YOLOv8
         <https://arxiv.org/pdf/2305.09972>} and from U{YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications
@@ -162,7 +162,7 @@ def decode_bbox(self, anchor_points: Tensor, pred_dist: Tensor) -> Tensor:
             )
         return dist2bbox(dist_transformed, anchor_points, out_format="xyxy")
 
-    def _init_parameters(self, features: list[Tensor]):
+    def _init_parameters(self, features: list[Tensor]) -> None:
         if not hasattr(self, "gt_bboxes_scale"):
             _, self.anchor_points, _, self.stride_tensor = (
                 anchors_for_fpn_features(

luxonis_train/attached_modules/losses/precision_dlf_segmentation_loss.py

@@ -1,5 +1,3 @@
-from typing import Any
-
 import torch
 import torch.nn.functional as F
 from torch import Tensor
@@ -22,7 +20,7 @@ def __init__(
         class_loss_weight: float = 0.5,
         bbox_loss_weight: float = 7.5,
         dfl_loss_weight: float = 1.5,
-        **kwargs: Any,
+        **kwargs,
     ):
         """Instance Segmentation and BBox loss adapted from  U{Real-Time Flying Object Detection with YOLOv8
         <https://arxiv.org/pdf/2305.09972>} and from U{YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications

luxonis_train/attached_modules/losses/reconstruction_segmentation_loss.py

@@ -118,7 +118,7 @@ def create_window(window_size: int, channel: int = 1) -> Tensor:
 
 
 def gaussian(window_size: int, sigma: float) -> Tensor:
-    gauss = torch.Tensor(
+    gauss = torch.tensor(
         [
             exp(-((x - window_size // 2) ** 2) / float(2 * sigma**2))
             for x in range(window_size)

luxonis_train/attached_modules/losses/sigmoid_focal_loss.py

@@ -1,4 +1,4 @@
-from typing import Any, Literal
+from typing import Literal
 
 from torch import Tensor
 from torchvision.ops import sigmoid_focal_loss
@@ -15,7 +15,7 @@ def __init__(
         alpha: float = 0.25,
         gamma: float = 2.0,
         reduction: Literal["none", "mean", "sum"] = "mean",
-        **kwargs: Any,
+        **kwargs,
     ):
         """Focal loss from U{Focal Loss for Dense Object Detection
         <https://arxiv.org/abs/1708.02002>}.

luxonis_train/attached_modules/losses/smooth_bce_with_logits.py

@@ -1,4 +1,4 @@
-from typing import Any, Literal
+from typing import Literal
 
 import torch
 from torch import Tensor
@@ -18,7 +18,7 @@ def __init__(
         bce_pow: float = 1.0,
         weight: list[float] | None = None,
         reduction: Literal["mean", "sum", "none"] = "mean",
-        **kwargs: Any,
+        **kwargs,
     ):
         """BCE with logits loss and label smoothing.
 

luxonis_train/attached_modules/losses/softmax_focal_loss.py

@@ -1,4 +1,4 @@
-from typing import Any, Literal
+from typing import Literal
 
 import torch
 from torch import Tensor
@@ -18,7 +18,7 @@ def __init__(
         gamma: float = 2.0,
         smooth: float = 0.0,
         reduction: Literal["none", "mean", "sum"] = "mean",
-        **kwargs: Any,
+        **kwargs,
     ):
         """Focal loss implementation for classification and segmentation
         tasks using Softmax.

luxonis_train/attached_modules/metrics/embedding_metrics.py

@@ -69,7 +69,7 @@ def update(self, predictions: Tensor, target: Tensor) -> None:
         self.correct_predictions += correct_predictions
         self.total_predictions += len(filtered_labels)
 
-    def compute(self):
+    def compute(self) -> Tensor:
         return self.correct_predictions / self.total_predictions
 
 
@@ -90,7 +90,7 @@ def __init__(self, **kwargs):
             "closest_vs_positive_distances", default=[], dist_reduce_fx="cat"
         )
 
-    def update(self, embeddings: Tensor, target: Tensor):
+    def update(self, embeddings: Tensor, target: Tensor) -> None:
         if self.cross_batch_memory_size is not None:
             self.cross_batch_memory.extend(list(zip(embeddings, target)))
 
@@ -137,7 +137,7 @@ def update(self, embeddings: Tensor, target: Tensor):
             closest_positive_distances[non_inf_mask]
         )
 
-    def compute(self):
+    def compute(self) -> dict[str, Tensor]:
         if len(self.all_distances) == 0:
             return {
                 "MedianDistance": torch.tensor(float("nan")),
@@ -165,18 +165,18 @@ def compute(self):
         }
 
 
-def _pairwise_distances(embeddings, squared=False):
+def _pairwise_distances(embeddings: Tensor, squared: bool = False) -> Tensor:
     """Compute the 2D matrix of distances between all the embeddings.
 
     @param embeddings: tensor of shape (batch_size, embed_dim)
-    @type embeddings: torch.Tensor
+    @type embeddings: Tensor
     @param squared: If true, output is the pairwise squared euclidean
         distance matrix. If false, output is the pairwise euclidean
         distance matrix.
     @type squared: bool
     @return: pairwise_distances: tensor of shape (batch_size,
         batch_size)
-    @rtype: torch.Tensor
+    @rtype: Tensor
     """
     dot_product = torch.matmul(embeddings, embeddings.t())
 
@@ -198,7 +198,7 @@ def _pairwise_distances(embeddings, squared=False):
     return distances
 
 
-def _get_anchor_positive_triplet_mask(labels):
+def _get_anchor_positive_triplet_mask(labels: Tensor) -> Tensor:
     indices_equal = torch.eye(
         labels.shape[0], dtype=torch.uint8, device=labels.device
     )

luxonis_train/attached_modules/metrics/mean_average_precision/mean_average_precision_bbox.py

@@ -1,5 +1,3 @@
-from typing import Any
-
 import torch
 import torchmetrics.detection as detection
 from torch import Tensor
@@ -20,7 +18,7 @@ class MeanAveragePrecisionBBox(BaseMetric, register=False):
 
     supported_tasks = [Tasks.BOUNDINGBOX]
 
-    def __init__(self, **kwargs: Any):
+    def __init__(self, **kwargs):
         super().__init__(**kwargs)
 
         self.metric = detection.MeanAveragePrecision(iou_type="bbox")

luxonis_train/attached_modules/metrics/object_keypoint_similarity.py

@@ -1,5 +1,4 @@
 from functools import cached_property
-from typing import Any
 
 import torch
 from scipy.optimize import linear_sum_assignment
@@ -35,7 +34,7 @@ def __init__(
         sigmas: list[float] | None = None,
         area_factor: float | None = None,
         use_cocoeval_oks: bool = True,
-        **kwargs: Any,
+        **kwargs,
     ) -> None:
         """Object Keypoint Similarity metric for evaluating keypoint
         predictions.

luxonis_train/attached_modules/metrics/ocr_accuracy.py

@@ -24,7 +24,7 @@ def __init__(self, blank_cls: int = 0, **kwargs):
         self.blank_cls = blank_cls
         self._init_metric()
 
-    def _init_metric(self):
+    def _init_metric(self) -> None:
         """Initializes the running metric."""
         self.running_metric = {
             "acc_0": 0.0,

luxonis_train/attached_modules/metrics/torchmetrics.py

@@ -1,6 +1,5 @@
 from contextlib import suppress
 from functools import cached_property
-from typing import Any
 
 import torchmetrics
 from torch import Tensor
@@ -14,7 +13,7 @@
 class TorchMetricWrapper(BaseMetric):
     Metric: type[torchmetrics.Metric]
 
-    def __init__(self, **kwargs: Any):
+    def __init__(self, **kwargs):
         super().__init__(node=kwargs.pop("node", None))
         task = kwargs.get("task")
         if task is None: