Revert D107142434: Enable Pyrefly in fbcode/vision/fair

generatedunixname89002005232357 · meta-codesync[bot] · commit 7684ebce0790 · 2026-06-02T16:19:57.000-07:00
Differential Revision:
D107142434

Original commit changeset: 25929bb3d5a3

Original Phabricator Diff: D107142434

fbshipit-source-id: 0aecaeba28d7d8db8f9273406a080e41aa77c4a7
diff --git a/projects/DensePose/densepose/converters/base.py b/projects/DensePose/densepose/converters/base.py
@@ -54,7 +54,6 @@ def _lookup_converter(cls, from_type: Type) -> Any:
                 if no suitable entry was found in the registry
         """
         if from_type in cls.registry:  # pyre-ignore[16]
-            # pyrefly: ignore [missing-attribute]
             return cls.registry[from_type]
         for base in from_type.__bases__:
             converter = cls._lookup_converter(base)
diff --git a/projects/DensePose/densepose/converters/chart_output_to_chart_result.py b/projects/DensePose/densepose/converters/chart_output_to_chart_result.py
@@ -93,7 +93,6 @@ def densepose_chart_predictor_output_to_result(
 
     boxes_xyxy_abs = boxes.tensor.clone()
     boxes_xywh_abs = BoxMode.convert(boxes_xyxy_abs, BoxMode.XYXY_ABS, BoxMode.XYWH_ABS)
-    # pyrefly: ignore [bad-argument-type]
     box_xywh = make_int_box(boxes_xywh_abs[0])
 
     labels = resample_fine_and_coarse_segm_to_bbox(predictor_output, box_xywh).squeeze(0)
@@ -184,7 +183,6 @@ def densepose_chart_predictor_output_to_result_with_confidences(
 
     boxes_xyxy_abs = boxes.tensor.clone()
     boxes_xywh_abs = BoxMode.convert(boxes_xyxy_abs, BoxMode.XYXY_ABS, BoxMode.XYWH_ABS)
-    # pyrefly: ignore [bad-argument-type]
     box_xywh = make_int_box(boxes_xywh_abs[0])
 
     labels = resample_fine_and_coarse_segm_to_bbox(predictor_output, box_xywh).squeeze(0)
diff --git a/projects/DensePose/densepose/converters/segm_to_mask.py b/projects/DensePose/densepose/converters/segm_to_mask.py
@@ -109,7 +109,6 @@ def predictor_output_with_coarse_segm_to_mask(
     N = len(boxes_xywh_abs)
     masks = torch.zeros((N, H, W), dtype=torch.bool, device=boxes.tensor.device)
     for i in range(len(boxes_xywh_abs)):
-        # pyrefly: ignore [bad-argument-type]
         box_xywh = make_int_box(boxes_xywh_abs[i])
         box_mask = resample_coarse_segm_tensor_to_bbox(predictor_output[i].coarse_segm, box_xywh)
         x, y, w, h = box_xywh
@@ -146,7 +145,6 @@ def predictor_output_with_fine_and_coarse_segm_to_mask(
     N = len(boxes_xywh_abs)
     masks = torch.zeros((N, H, W), dtype=torch.bool, device=boxes.tensor.device)
     for i in range(len(boxes_xywh_abs)):
-        # pyrefly: ignore [bad-argument-type]
         box_xywh = make_int_box(boxes_xywh_abs[i])
         labels_i = resample_fine_and_coarse_segm_to_bbox(predictor_output[i], box_xywh)
         x, y, w, h = box_xywh
diff --git a/projects/DensePose/densepose/data/build.py b/projects/DensePose/densepose/data/build.py
@@ -225,7 +225,6 @@ def _maybe_create_specific_keep_instance_predicate(cfg: CfgNode) -> Optional[Ins
         return None
 
     def combined_predicate(instance: Instance) -> bool:
-        # pyrefly: ignore [not-callable]
         return any(p(instance) for p in predicates)
 
     return combined_predicate
@@ -236,7 +235,6 @@ def _get_train_keep_instance_predicate(cfg: CfgNode):
     combined_specific_keep_predicate = _maybe_create_specific_keep_instance_predicate(cfg)
 
     def combined_general_specific_keep_predicate(instance: Instance) -> bool:
-        # pyrefly: ignore [not-callable]
         return general_keep_predicate(instance) and combined_specific_keep_predicate(instance)
 
     if (general_keep_predicate is None) and (combined_specific_keep_predicate is None):
@@ -388,9 +386,7 @@ def combine_detection_dataset_dicts(
     """
     assert len(dataset_names)
     if proposal_files is None:
-        # pyrefly: ignore [bad-assignment]
         proposal_files = [None] * len(dataset_names)
-    # pyrefly: ignore [bad-argument-type]
     assert len(dataset_names) == len(proposal_files)
     # load datasets and metadata
     dataset_name_to_dicts = {}
@@ -407,7 +403,6 @@ def combine_detection_dataset_dicts(
     # map to contiguous category IDs
     _add_category_id_to_contiguous_id_maps_to_metadata(merged_categories)
     # load annotations and dataset metadata
-    # pyrefly: ignore [bad-argument-type]
     for dataset_name, proposal_file in zip(dataset_names, proposal_files):
         dataset_dicts = dataset_name_to_dicts[dataset_name]
         assert len(dataset_dicts), f"Dataset '{dataset_name}' is empty!"
@@ -552,7 +547,6 @@ def build_bootstrap_dataset(dataset_name: str, cfg: CfgNode) -> Sequence[torch.T
         dataset = factory(meta, cfg)
     if dataset is None:
         logger.warning(f"Failed to create dataset {dataset_name} of type {meta.dataset_type}")
-    # pyrefly: ignore [bad-return]
     return dataset
 
 
@@ -741,5 +735,4 @@ def register(self, dataset_type: DatasetType, factory: Callable[[Metadata, CfgNo
 
 
 BootstrapDatasetFactoryCatalog = _BootstrapDatasetFactoryCatalog()
-# pyrefly: ignore [bad-argument-type]
 BootstrapDatasetFactoryCatalog.register(DatasetType.VIDEO_LIST, build_video_list_dataset)
diff --git a/projects/DensePose/densepose/data/dataset_mapper.py b/projects/DensePose/densepose/data/dataset_mapper.py
@@ -157,7 +157,6 @@ def _add_densepose_masks_as_segmentation(
             bbox_segm_dp = torch.tensor((0, 0, segm_h - 1, segm_w - 1), dtype=torch.float32)
             # image bbox
             x0, y0, x1, y1 = (
-                # pyrefly: ignore [missing-attribute]
                 v.item() for v in BoxMode.convert(obj["bbox"], obj["bbox_mode"], BoxMode.XYXY_ABS)
             )
             segm_aligned = (
diff --git a/projects/DensePose/densepose/data/datasets/coco.py b/projects/DensePose/densepose/data/datasets/coco.py
@@ -253,7 +253,6 @@ def _combine_images_with_annotations(
         record["dataset"] = dataset_name
         if "frame_id" in img_dict:
             record["frame_id"] = img_dict["frame_id"]
-            # pyrefly: ignore [unsupported-operation]
             record["video_id"] = img_dict.get("vid_id", None)
             contains_video_frame_info = True
         objs = []
@@ -266,7 +265,6 @@ def _combine_images_with_annotations(
             _maybe_add_keypoints(obj, ann_dict)
             _maybe_add_densepose(obj, ann_dict)
             objs.append(obj)
-        # pyrefly: ignore [unsupported-operation]
         record["annotations"] = objs
         dataset_dicts.append(record)
     if contains_video_frame_info:
diff --git a/projects/DensePose/densepose/data/image_list_dataset.py b/projects/DensePose/densepose/data/image_list_dataset.py
@@ -36,15 +36,13 @@ def __init__(
         if type(category_list) is list:
             self.category_list = category_list
         else:
-            # pyrefly: ignore [bad-assignment]
             self.category_list = [category_list] * len(image_list)
         assert len(image_list) == len(
             self.category_list
         ), "length of image and category lists must be equal"
         self.image_list = image_list
         self.transform = transform
 
-    # pyrefly: ignore [bad-override-param-name]
     def __getitem__(self, idx: int) -> Dict[str, Any]:
         """
         Gets selected images from the list
diff --git a/projects/DensePose/densepose/data/meshes/catalog.py b/projects/DensePose/densepose/data/meshes/catalog.py
@@ -26,7 +26,6 @@ def __init__(self, *args, **kwargs):
         self.mesh_names = {}
         self.max_mesh_id = -1
 
-    # pyrefly: ignore [bad-override-param-name]
     def __setitem__(self, key, value):
         if key in self:
             logger = logging.getLogger(__name__)
diff --git a/projects/DensePose/densepose/data/samplers/densepose_base.py b/projects/DensePose/densepose/data/samplers/densepose_base.py
@@ -39,7 +39,6 @@ def __call__(self, instances: Instances) -> DensePoseList:
         boxes_xywh_abs = BoxMode.convert(boxes_xyxy_abs, BoxMode.XYXY_ABS, BoxMode.XYWH_ABS)
         dp_datas = []
         for i in range(len(boxes_xywh_abs)):
-            # pyrefly: ignore [bad-argument-type]
             annotation_i = self._sample(instances[i], make_int_box(boxes_xywh_abs[i]))
             annotation_i[DensePoseDataRelative.S_KEY] = self._resample_mask(  # pyre-ignore[6]
                 instances[i].pred_densepose
diff --git a/projects/DensePose/densepose/data/samplers/densepose_cse_base.py b/projects/DensePose/densepose/data/samplers/densepose_cse_base.py
@@ -70,7 +70,6 @@ def _sample(self, instance: Instances, bbox_xywh: IntTupleBox) -> Dict[str, List
 
         count = min(self.count_per_class, k)
         if count <= 0:
-            # pyrefly: ignore [bad-return]
             return annotation
 
         index_sample = self._produce_index_sample(values, count)
@@ -86,13 +85,9 @@ def _sample(self, instance: Instances, bbox_xywh: IntTupleBox) -> Dict[str, List
         x = (sampled_x / w * 256.0).cpu().tolist()
         y = (sampled_y / h * 256.0).cpu().tolist()
         # extend annotations
-        # pyrefly: ignore [missing-attribute]
         annotation[DensePoseDataRelative.X_KEY].extend(x)
-        # pyrefly: ignore [missing-attribute]
         annotation[DensePoseDataRelative.Y_KEY].extend(y)
-        # pyrefly: ignore [missing-attribute]
         annotation[DensePoseDataRelative.VERTEX_IDS_KEY].extend(closest_vertices.cpu().tolist())
-        # pyrefly: ignore [bad-return]
         return annotation
 
     def _produce_mask_and_results(
diff --git a/projects/DensePose/densepose/data/video/video_keyframe_dataset.py b/projects/DensePose/densepose/data/video/video_keyframe_dataset.py
@@ -254,7 +254,6 @@ def __init__(
         if type(category_list) is list:
             self.category_list = category_list
         else:
-            # pyrefly: ignore [bad-assignment]
             self.category_list = [category_list] * len(video_list)
         assert len(video_list) == len(
             self.category_list
@@ -268,7 +267,6 @@ def __init__(
             else None
         )
 
-    # pyrefly: ignore [bad-override-param-name]
     def __getitem__(self, idx: int) -> Dict[str, Any]:
         """
         Gets selected keyframes from a given video
diff --git a/projects/DensePose/densepose/engine/trainer.py b/projects/DensePose/densepose/engine/trainer.py
@@ -79,7 +79,6 @@ def extract_embedder_from_model(cls, model: nn.Module) -> Optional[Embedder]:
         if isinstance(model, nn.parallel.DistributedDataParallel):
             model = model.module
         if hasattr(model, "roi_heads") and hasattr(model.roi_heads, "embedder"):
-            # pyrefly: ignore [bad-return]
             return model.roi_heads.embedder
         return None
 
@@ -175,7 +174,6 @@ def build_evaluator(
         if cfg.MODEL.DENSEPOSE_ON:
             storage = build_densepose_evaluator_storage(cfg, output_folder)
             evaluators.append(
-                # pyrefly: ignore [bad-argument-type]
                 DensePoseCOCOEvaluator(
                     dataset_name,
                     distributed,
diff --git a/projects/DensePose/densepose/evaluation/densepose_coco_evaluation.py b/projects/DensePose/densepose/evaluation/densepose_coco_evaluation.py
@@ -600,9 +600,7 @@ def _extract_iuv(
         elif self._dpDataMode == DensePoseDataMode.I_GT_UV_0:
             # ground truth labels, UV = 0
             ipoints = np.array(gt["dp_I"])
-            # pyrefly: ignore [unbound-name]
             upoints = upoints * 0.0
-            # pyrefly: ignore [unbound-name]
             vpoints = vpoints * 0.0
         elif self._dpDataMode == DensePoseDataMode.I_GT_UV_DT:
             # ground truth labels, estimated UV
@@ -612,9 +610,7 @@ def _extract_iuv(
         elif self._dpDataMode == DensePoseDataMode.I_DT_UV_0:
             # estimated labels, UV = 0
             ipoints = densepose_data[0, py, px]
-            # pyrefly: ignore [unbound-name]
             upoints = upoints * 0.0
-            # pyrefly: ignore [unbound-name]
             vpoints = vpoints * 0.0
         else:
             raise ValueError(f"Unknown data mode: {self._dpDataMode}")
diff --git a/projects/DensePose/densepose/evaluation/evaluator.py b/projects/DensePose/densepose/evaluation/evaluator.py
@@ -69,7 +69,6 @@ def __init__(
         ), "Mesh alignment evaluation is activated, but no vertex embedder provided!"
         if should_evaluate_mesh_alignment:
             self._mesh_alignment_evaluator = MeshAlignmentEvaluator(
-                # pyrefly: ignore [bad-argument-type]
                 embedder,
                 mesh_alignment_mesh_names,
             )
@@ -399,21 +398,15 @@ def build_densepose_evaluator_storage(cfg: CfgNode, output_folder: str):
     if evaluator_type == "iuv":
         n_fsc = cfg.MODEL.ROI_DENSEPOSE_HEAD.NUM_PATCHES + 1
         schema = {
-            # pyrefly: ignore [bad-argument-type]
             "coarse_segm": SizeData(dtype="float32", shape=(n_csc, hout, wout)),
-            # pyrefly: ignore [bad-argument-type]
             "fine_segm": SizeData(dtype="float32", shape=(n_fsc, hout, wout)),
-            # pyrefly: ignore [bad-argument-type]
             "u": SizeData(dtype="float32", shape=(n_fsc, hout, wout)),
-            # pyrefly: ignore [bad-argument-type]
             "v": SizeData(dtype="float32", shape=(n_fsc, hout, wout)),
         }
     elif evaluator_type == "cse":
         embed_size = cfg.MODEL.ROI_DENSEPOSE_HEAD.CSE.EMBED_SIZE
         schema = {
-            # pyrefly: ignore [bad-argument-type]
             "coarse_segm": SizeData(dtype="float32", shape=(n_csc, hout, wout)),
-            # pyrefly: ignore [bad-argument-type]
             "embedding": SizeData(dtype="float32", shape=(embed_size, hout, wout)),
         }
     else:
diff --git a/projects/DensePose/densepose/modeling/hrfpn.py b/projects/DensePose/densepose/modeling/hrfpn.py
@@ -134,7 +134,6 @@ def init_weights(self):
                 nn.init.kaiming_normal_(m.weight, a=1)
                 nn.init.constant_(m.bias, 0)
 
-    # pyrefly: ignore [bad-override]
     def forward(self, inputs):
         bottom_up_features = self.bottom_up(inputs)
         assert len(bottom_up_features) == len(self.in_features)
diff --git a/projects/DensePose/densepose/modeling/hrnet.py b/projects/DensePose/densepose/modeling/hrnet.py
@@ -433,7 +433,6 @@ def _make_stage(self, layer_config, num_inchannels, multi_scale_output=True):
 
         return nn.Sequential(*modules), num_inchannels
 
-    # pyrefly: ignore [bad-override]
     def forward(self, x):
         x = self.conv1(x)
         x = self.bn1(x)
diff --git a/projects/DensePose/densepose/modeling/losses/chart_with_confidences.py b/projects/DensePose/densepose/modeling/losses/chart_with_confidences.py
@@ -28,7 +28,6 @@ def __init__(self, cfg: CfgNode):
                 self.confidence_model_cfg.uv_confidence.epsilon
             )
         elif self.confidence_model_cfg.uv_confidence.type == DensePoseUVConfidenceType.INDEP_ANISO:
-            # pyrefly: ignore [bad-assignment]
             self.uv_loss_with_confidences = IndepAnisotropicGaussianUVLoss(
                 self.confidence_model_cfg.uv_confidence.epsilon
             )
diff --git a/projects/DensePose/densepose/modeling/losses/embed.py b/projects/DensePose/densepose/modeling/losses/embed.py
@@ -111,7 +111,6 @@ def __call__(
                 losses[mesh_name] = self.fake_value(
                     densepose_predictor_outputs, embedder, mesh_name
                 )
-        # pyrefly: ignore [bad-return]
         return losses
 
     def fake_values(self, densepose_predictor_outputs: Any, embedder: nn.Module):
diff --git a/projects/DensePose/densepose/modeling/losses/soft_embed.py b/projects/DensePose/densepose/modeling/losses/soft_embed.py
@@ -125,7 +125,6 @@ def __call__(
                 losses[mesh_name] = self.fake_value(
                     densepose_predictor_outputs, embedder, mesh_name
                 )
-        # pyrefly: ignore [bad-return]
         return losses
 
     def fake_values(self, densepose_predictor_outputs: Any, embedder: nn.Module):
diff --git a/projects/DensePose/densepose/modeling/predictors/chart_confidence.py b/projects/DensePose/densepose/modeling/predictors/chart_confidence.py
@@ -76,7 +76,6 @@ def _initialize_confidence_estimation_layers(self, cfg: CfgNode, dim_in: int):
                 )
             else:
                 raise ValueError(
-                    # pyrefly: ignore [missing-attribute]
                     f"Unknown confidence model type: "
                     f"{self.confidence_model_cfg.confidence_model_type}"
                 )
@@ -115,13 +114,11 @@ def forward(self, head_outputs: torch.Tensor):
                 == DensePoseUVConfidenceType.INDEP_ANISO
             ):
                 # assuming base class defines interp2d method for bilinear interpolation
-                # pyrefly: ignore [missing-attribute]
                 output.sigma_2 = self.interp2d(self.sigma_2_lowres(head_outputs))
                 output.kappa_u = self.interp2d(self.kappa_u_lowres(head_outputs))  # pyre-ignore[16]
                 output.kappa_v = self.interp2d(self.kappa_v_lowres(head_outputs))  # pyre-ignore[16]
             else:
                 raise ValueError(
-                    # pyrefly: ignore [missing-attribute]
                     f"Unknown confidence model type: "
                     f"{self.confidence_model_cfg.confidence_model_type}"
                 )
@@ -139,7 +136,6 @@ def forward(self, head_outputs: torch.Tensor):
             )
             output.coarse_segm_confidence = (
                 F.softplus(
-                    # pyrefly: ignore [missing-attribute]
                     self.interp2d(
                         self.coarse_segm_confidence_lowres(head_outputs)  # pyre-ignore[16]
                     )
diff --git a/projects/DensePose/densepose/modeling/roi_heads/deeplab.py b/projects/DensePose/densepose/modeling/roi_heads/deeplab.py
@@ -101,7 +101,6 @@ def __init__(self, in_channels, out_channels):
             nn.ReLU(),
         )
 
-    # pyrefly: ignore [bad-override-param-name]
     def forward(self, x):
         size = x.shape[-2:]
         x = super(ASPPPooling, self).forward(x)
diff --git a/projects/DensePose/densepose/modeling/roi_heads/roi_head.py b/projects/DensePose/densepose/modeling/roi_heads/roi_head.py
@@ -69,7 +69,6 @@ def __init__(self, cfg, input_shape: Dict[str, ShapeSpec], in_features):
                 head_ops.append(conv)
                 if feature_strides[in_feature] != self.common_stride:
                     head_ops.append(
-                        # pyrefly: ignore [bad-argument-type]
                         nn.Upsample(scale_factor=2, mode="bilinear", align_corners=False)
                     )
             self.scale_heads.append(nn.Sequential(*head_ops))
@@ -82,9 +81,7 @@ def forward(self, features: List[torch.Tensor]):
             if i == 0:
                 x = self.scale_heads[i](features[i])
             else:
-                # pyrefly: ignore [unbound-name]
                 x = x + self.scale_heads[i](features[i])
-        # pyrefly: ignore [unbound-name]
         x = self.predictor(x)
         return x
 
diff --git a/projects/DensePose/densepose/structures/chart_result.py b/projects/DensePose/densepose/structures/chart_result.py
@@ -158,7 +158,6 @@ def compress_quantized_densepose_chart_result(
     im.save(fstream, format="png", optimize=True)
     labels_uv_str = base64.encodebytes(fstream.getvalue()).decode()
     shape_chw = labels_uv_uint8_np_chw.shape
-    # pyrefly: ignore [bad-argument-type]
     return DensePoseChartResultCompressed(labels_uv_str=labels_uv_str, shape_chw=shape_chw)
 
 
diff --git a/projects/DensePose/densepose/structures/mesh.py b/projects/DensePose/densepose/structures/mesh.py
@@ -140,7 +140,6 @@ def load_mesh_data(
         # pyre-fixme[7]: Expected `Tuple[Optional[Tensor], Optional[Tensor]]` but
         #  got `Tensor`.
         return torch.as_tensor(pickle.load(hFile)[field], dtype=torch.float).to(device)
-    # pyrefly: ignore [bad-return]
     return None
 
 
diff --git a/projects/DensePose/densepose/structures/transform_data.py b/projects/DensePose/densepose/structures/transform_data.py
@@ -68,8 +68,6 @@ def load(io: Union[str, BinaryIO]):
             map_dst = uv_symmetry_map_torch[key]
             for i in range(map_src.shape[1]):
                 map_dst.append(torch.from_numpy(map_src[0, i]).to(dtype=torch.float))
-            # pyrefly: ignore [unsupported-operation]
             uv_symmetry_map_torch[key] = torch.stack(map_dst, dim=0)
-        # pyrefly: ignore [bad-argument-type]
         transform_data = DensePoseTransformData(uv_symmetry_map_torch, device=torch.device("cpu"))
         return transform_data
diff --git a/projects/DensePose/densepose/vis/densepose_results.py b/projects/DensePose/densepose/vis/densepose_results.py
@@ -92,7 +92,6 @@ def __init__(self, levels=10, **kwargs):
         self.levels = levels
         self.plot_args = kwargs
 
-    # pyrefly: ignore [bad-override]
     def create_visualization_context(self, image_bgr: Image):
         import matplotlib.pyplot as plt
         from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
@@ -107,7 +106,6 @@ def create_visualization_context(self, image_bgr: Image):
         plt.axis("off")
         context["fig"] = fig
         canvas = FigureCanvas(fig)
-        # pyrefly: ignore [unsupported-operation]
         context["canvas"] = canvas
         extent = (0, image_bgr.shape[1], image_bgr.shape[0], 0)
         plt.imshow(image_bgr[:, :, ::-1], extent=extent)
diff --git a/projects/DensePose/densepose/vis/extractor.py b/projects/DensePose/densepose/vis/extractor.py
@@ -172,9 +172,7 @@ def __call__(self, instances: Instances, select=None):
             return None
         select_local_idx = batched_nms(
             boxes_xywh,
-            # pyrefly: ignore [bad-argument-type]
             scores,
-            # pyrefly: ignore [bad-argument-type]
             torch.zeros(len(scores), dtype=torch.int32),
             iou_threshold=self.iou_threshold,
         ).squeeze()

Original file line number	Diff line number	Diff line change
`@@ -157,7 +157,6 @@ def _add_densepose_masks_as_segmentation(`
`157`	`157`	`bbox_segm_dp = torch.tensor((0, 0, segm_h - 1, segm_w - 1), dtype=torch.float32)`
`158`	`158`	`# image bbox`
`159`	`159`	`x0, y0, x1, y1 = (`
`160`		`- # pyrefly: ignore [missing-attribute]`
`161`	`160`	`v.item() for v in BoxMode.convert(obj["bbox"], obj["bbox_mode"], BoxMode.XYXY_ABS)`
`162`	`161`	`)`
`163`	`162`	`segm_aligned = (`