apply Black 25.11.0 style in fbcode/deeplearning/projects (21/92)

Summary:
Formats the covered files with pyfmt.

paintitblack

Reviewed By: itamaro

Differential Revision: D90476315

fbshipit-source-id: ee94c471788b8e7d067813d8b3e0311214d17f3f

sam3/train/transforms/basic.py

@@ -14,7 +14,6 @@ import PIL
 import torch
 import torchvision.transforms as T
 import torchvision.transforms.functional as F
-
 from sam3.model.box_ops import box_xyxy_to_cxcywh
 from sam3.model.data_misc import interpolate
 
@@ -277,9 +276,9 @@ class RandomSizeCrop:
                     max(0, minY - h + 1), max(maxY - 1, max(0, minY - h + 1))
                 )
             result_img, result_target = crop(img, target, [j, i, h, w])
-            assert (
-                len(result_target["boxes"]) == init_boxes
-            ), f"img_w={img.width}\timg_h={img.height}\tminX={minX}\tminY={minY}\tmaxX={maxX}\tmaxY={maxY}\tminW={minW}\tminH={minH}\tmaxW={maxW}\tmaxH={maxH}\tw={w}\th={h}\ti={i}\tj={j}\tinit_boxes={init_boxes_tensor}\tresults={result_target['boxes']}"
+            assert len(result_target["boxes"]) == init_boxes, (
+                f"img_w={img.width}\timg_h={img.height}\tminX={minX}\tminY={minY}\tmaxX={maxX}\tmaxY={maxY}\tminW={minW}\tminH={minH}\tmaxW={maxW}\tmaxH={maxH}\tw={w}\th={h}\ti={i}\tj={j}\tinit_boxes={init_boxes_tensor}\tresults={result_target['boxes']}"
+            )
 
             return result_img, result_target
         else:

sam3/train/transforms/basic_for_api.py

@@ -7,7 +7,6 @@ Transforms and data augmentation for both image + bbox.
 """
 
 import logging
-
 import numbers
 import random
 from collections.abc import Sequence
@@ -17,9 +16,7 @@ import torch
 import torchvision.transforms as T
 import torchvision.transforms.functional as F
 import torchvision.transforms.v2.functional as Fv2
-
 from PIL import Image as PILImage
-
 from sam3.model.box_ops import box_xyxy_to_cxcywh, masks_to_boxes
 from sam3.train.data.sam3_image_dataset import Datapoint
 from torchvision.transforms import InterpolationMode

sam3/train/transforms/filter_query_transforms.py

@@ -4,12 +4,10 @@
 
 import logging
 import random
-
 from collections import defaultdict
 from typing import List, Optional, Union
 
 import torch
-
 from sam3.train.data.sam3_image_dataset import Datapoint, FindQuery, Object
 
 
@@ -381,9 +379,9 @@ class FlexibleFilterFindGetQueries:
         if len(new_find_queries) == 0:
             start_with_zero_check = True
 
-        assert (
-            start_with_zero_check
-        ), "Invalid Find queries, they need to start at query_processing_order = 0"
+        assert start_with_zero_check, (
+            "Invalid Find queries, they need to start at query_processing_order = 0"
+        )
 
         datapoint.find_queries = new_find_queries
 

sam3/train/transforms/point_sampling.py

@@ -7,7 +7,6 @@ import numpy as np
 import torch
 from PIL import Image as PILImage
 from pycocotools import mask as mask_util
-
 from sam3.train.data.sam3_image_dataset import Datapoint
 from torchvision.ops import masks_to_boxes
 
@@ -250,9 +249,9 @@ class RandomGeometricInputsAPI:
     def _get_target_object(self, datapoint, query):
         img = datapoint.images[query.image_id]
         targets = query.object_ids_output
-        assert (
-            len(targets) == 1
-        ), "Geometric queries only support a single target object."
+        assert len(targets) == 1, (
+            "Geometric queries only support a single target object."
+        )
         target_idx = targets[0]
         return img.objects[target_idx]
 

sam3/train/transforms/segmentation.py

@@ -5,12 +5,9 @@
 import numpy as np
 import pycocotools.mask as mask_utils
 import torch
-
 import torchvision.transforms.functional as F
 from PIL import Image as PILImage
-
 from sam3.model.box_ops import masks_to_boxes
-
 from sam3.train.data.sam3_image_dataset import Datapoint