suppress errors in `vision/fair/pytorch3d`

Differential Revision: D37172764

fbshipit-source-id: a2ec367e56de2781a17f5e708eb5832ec9d7e6b4

pytorch3d/implicitron/tools/circle_fitting.py

@@ -108,6 +108,7 @@ def fit_circle_in_2d(
         raise ValueError(f"{n_provided} points are not enough to determine a circle")
     solution = lstsq(design, rhs[:, None])
     center = solution[:2, 0] / 2
+    # pyre-fixme[58]: `**` is not supported for operand types `Tensor` and `int`.
     radius = torch.sqrt(solution[2, 0] + (center**2).sum())
     if n_points > 0:
         if angles is not None:

pytorch3d/implicitron/tools/depth_cleanup.py

@@ -50,6 +50,7 @@ def cleanup_eval_depth(
     # the threshold is a sigma-multiple of the standard deviation of the depth
     mu = wmean(depth.view(ba, -1, 1), mask.view(ba, -1)).view(ba, 1)
     std = (
+        # pyre-fixme[58]: `**` is not supported for operand types `Tensor` and `int`.
         wmean((depth.view(ba, -1) - mu).view(ba, -1, 1) ** 2, mask.view(ba, -1))
         .clamp(1e-4)
         .sqrt()
@@ -62,7 +63,6 @@ def cleanup_eval_depth(
     # print(f'Kept {100.0 * perc_kept.mean():1.3f} % points')
 
     good_depth_raster = torch.zeros_like(depth).view(ba, -1)
-    # pyre-ignore[16]: scatter_add_
     good_depth_raster.scatter_add_(1, torch.round(idx_sampled[:, 0]).long(), good_depth)
 
     good_depth_mask = (good_depth_raster.view(ba, 1, H, W) > 0).float()

pytorch3d/implicitron/tools/metric_utils.py

@@ -65,6 +65,7 @@ def eval_depth(
 
     df = gt - pred
 
+    # pyre-fixme[58]: `**` is not supported for operand types `Tensor` and `int`.
     mse_depth = (dmask * (df**2)).sum((1, 2, 3)) / dmask_mass
     abs_depth = (dmask * df.abs()).sum((1, 2, 3)) / dmask_mass
 
@@ -100,8 +101,10 @@ def calc_mse(
     Calculates the mean square error between tensors `x` and `y`.
     """
     if mask is None:
+        # pyre-fixme[58]: `**` is not supported for operand types `Tensor` and `int`.
         return torch.mean((x - y) ** 2)
     else:
+        # pyre-fixme[58]: `**` is not supported for operand types `Tensor` and `int`.
         return (((x - y) ** 2) * mask).sum() / mask.expand_as(x).sum().clamp(1e-5)
 
 
@@ -128,6 +131,7 @@ def calc_bce(
         mask_bg = (1 - mask_fg) * mask
         weight = mask_fg / mask_fg.sum().clamp(1.0) + mask_bg / mask_bg.sum().clamp(1.0)
         # weight sum should be at this point ~2
+        # pyre-fixme[58]: `/` is not supported for operand types `int` and `Tensor`.
         weight = weight * (weight.numel() / weight.sum().clamp(1.0))
     else:
         weight = torch.ones_like(gt) * mask

pytorch3d/implicitron/tools/point_cloud_utils.py

@@ -55,8 +55,6 @@ def get_rgbd_point_cloud(
 
     pts_colors = torch.nn.functional.interpolate(
         image_rgb,
-        # pyre-fixme[6]: Expected `Optional[int]` for 2nd param but got
-        #  `List[typing.Any]`.
         size=[imh, imw],
         mode="bilinear",
         align_corners=False,
@@ -133,6 +131,7 @@ def render_point_cloud_pytorch3d(
     cumprod = torch.cat((torch.ones_like(cumprod[..., :1]), cumprod[..., :-1]), dim=-1)
     depths = (weights * cumprod * fragments.zbuf).sum(dim=-1)
     # add the rendering mask
+    # pyre-fixme[6]: For 1st param expected `Tensor` but got `float`.
     render_mask = -torch.prod(1.0 - weights, dim=-1) + 1.0
 
     # cat depths and render mask
@@ -141,8 +140,6 @@ def render_point_cloud_pytorch3d(
     # reshape back
     rendered_blob = Fu.interpolate(
         rendered_blob,
-        # pyre-fixme[6]: Expected `Optional[int]` for 2nd param but got `Tuple[int,
-        #  ...]`.
         size=tuple(render_size),
         mode="bilinear",
     )

pytorch3d/implicitron/tools/vis_utils.py

@@ -99,8 +99,6 @@ def visualize_basics(
             v = v.repeat(1, 3, 1, 1)
         v = torch.nn.functional.interpolate(
             v,
-            # pyre-fixme[6]: Expected `Optional[typing.List[float]]` for 2nd param
-            #  but got `float`.
             scale_factor=(
                 600.0
                 if (

pytorch3d/io/mtl_io.py

@@ -288,7 +288,6 @@ def make_material_atlas(
     # w0, w1
     bary[below_diag, slc] = ((grid[below_diag] + 1.0 / 3.0) / R).T
     # w0, w1 for above diagonal grid cells.
-    # pyre-fixme[16]: `float` has no attribute `T`.
     bary[~below_diag, slc] = (((R - 1.0 - grid[~below_diag]) + 2.0 / 3.0) / R).T
     # w2 = 1. - w0 - w1
     bary[..., -1] = 1 - bary[..., :2].sum(dim=-1)

pytorch3d/io/obj_io.py

@@ -57,7 +57,6 @@ def _format_faces_indices(faces_indices, max_index: int, device, pad_value=None)
     )
 
     if pad_value is not None:
-        # pyre-fixme[28]: Unexpected keyword argument `dim`.
         mask = faces_indices.eq(pad_value).all(dim=-1)
 
     # Change to 0 based indexing.

pytorch3d/io/utils.py

@@ -58,7 +58,6 @@ def _check_faces_indices(
     if pad_value is None:
         mask = torch.ones(faces_indices.shape[:-1]).bool()  # Keep all faces
     else:
-        # pyre-fixme[16]: `torch.ByteTensor` has no attribute `any`
         mask = faces_indices.ne(pad_value).any(dim=-1)
     if torch.any(faces_indices[mask] >= max_index) or torch.any(
         faces_indices[mask] < 0

pytorch3d/loss/mesh_laplacian_smoothing.py

@@ -112,6 +112,8 @@ def mesh_laplacian_smoothing(meshes, method: str = "uniform"):
             if method == "cot":
                 norm_w = torch.sparse.sum(L, dim=1).to_dense().view(-1, 1)
                 idx = norm_w > 0
+                # pyre-fixme[58]: `/` is not supported for operand types `float` and
+                #  `Tensor`.
                 norm_w[idx] = 1.0 / norm_w[idx]
             else:
                 L_sum = torch.sparse.sum(L, dim=1).to_dense().view(-1, 1)

pytorch3d/loss/point_mesh_distance.py

@@ -303,8 +303,8 @@ def point_mesh_edge_distance(meshes: Meshes, pcls: Pointclouds):
     # weight each example by the inverse of number of points in the example
     point_to_cloud_idx = pcls.packed_to_cloud_idx()  # (sum(P_i), )
     num_points_per_cloud = pcls.num_points_per_cloud()  # (N,)
-    # pyre-ignore[16]: `torch.Tensor` has no attribute `gather`
     weights_p = num_points_per_cloud.gather(0, point_to_cloud_idx)
+    # pyre-fixme[58]: `/` is not supported for operand types `float` and `Tensor`.
     weights_p = 1.0 / weights_p.float()
     point_to_edge = point_to_edge * weights_p
     point_dist = point_to_edge.sum() / N
@@ -378,8 +378,8 @@ def point_mesh_face_distance(
     # weight each example by the inverse of number of points in the example
     point_to_cloud_idx = pcls.packed_to_cloud_idx()  # (sum(P_i),)
     num_points_per_cloud = pcls.num_points_per_cloud()  # (N,)
-    # pyre-ignore[16]: `torch.Tensor` has no attribute `gather`
     weights_p = num_points_per_cloud.gather(0, point_to_cloud_idx)
+    # pyre-fixme[58]: `/` is not supported for operand types `float` and `Tensor`.
     weights_p = 1.0 / weights_p.float()
     point_to_face = point_to_face * weights_p
     point_dist = point_to_face.sum() / N

pytorch3d/ops/cameras_alignment.py

@@ -119,11 +119,16 @@ def corresponding_cameras_alignment(
 
     # create a new cameras object and set the R and T accordingly
     cameras_src_aligned = cameras_src.clone()
+    # pyre-fixme[6]: For 2nd param expected `Tensor` but got `Union[Tensor, Module]`.
     cameras_src_aligned.R = torch.bmm(align_t_R.expand_as(cameras_src.R), cameras_src.R)
     cameras_src_aligned.T = (
         torch.bmm(
-            align_t_T[:, None].repeat(cameras_src.R.shape[0], 1, 1), cameras_src.R
+            align_t_T[:, None].repeat(cameras_src.R.shape[0], 1, 1),
+            # pyre-fixme[6]: For 2nd param expected `Tensor` but got `Union[Tensor,
+            #  Module]`.
+            cameras_src.R,
         )[:, 0]
+        # pyre-fixme[29]: `Union[BoundMethod[typing.Callable(torch._C._TensorBase.__m...
         + cameras_src.T * align_t_s
     )
 
@@ -171,6 +176,7 @@ def _align_camera_extrinsics(
         R_A = (U V^T)^T
         ```
     """
+    # pyre-fixme[6]: For 1st param expected `Tensor` but got `Union[Tensor, Module]`.
     RRcov = torch.bmm(cameras_src.R, cameras_tgt.R.transpose(2, 1)).mean(0)
     U, _, V = torch.svd(RRcov)
     align_t_R = V @ U.t()
@@ -204,11 +210,13 @@ def _align_camera_extrinsics(
     #  `Union[BoundMethod[typing.Callable(torch.Tensor.__getitem__)[[Named(self,
     #  torch.Tensor), Named(item, typing.Any)], typing.Any], torch.Tensor],
     #  torch.Tensor, torch.nn.Module]` is not a function.
+    # pyre-fixme[6]: For 1st param expected `Tensor` but got `Union[Tensor, Module]`.
     A = torch.bmm(cameras_src.R, cameras_src.T[:, :, None])[:, :, 0]
     # pyre-fixme[29]:
     #  `Union[BoundMethod[typing.Callable(torch.Tensor.__getitem__)[[Named(self,
     #  torch.Tensor), Named(item, typing.Any)], typing.Any], torch.Tensor],
     #  torch.Tensor, torch.nn.Module]` is not a function.
+    # pyre-fixme[6]: For 1st param expected `Tensor` but got `Union[Tensor, Module]`.
     B = torch.bmm(cameras_src.R, cameras_tgt.T[:, :, None])[:, :, 0]
     Amu = A.mean(0, keepdim=True)
     Bmu = B.mean(0, keepdim=True)
@@ -217,6 +225,7 @@ def _align_camera_extrinsics(
         # of centered A and centered B
         Ac = A - Amu
         Bc = B - Bmu
+        # pyre-fixme[58]: `**` is not supported for operand types `Tensor` and `int`.
         align_t_s = (Ac * Bc).mean() / (Ac**2).mean().clamp(eps)
     else:
         # set the scale to identity

pytorch3d/ops/cubify.py

@@ -235,7 +235,6 @@ def cubify(voxels, thresh, device=None, align: str = "topleft") -> Meshes:
     idlenum = idleverts.cumsum(1)
 
     verts_list = [
-        # pyre-fixme[16]: `Tensor` has no attribute `index_select`.
         grid_verts.index_select(0, (idleverts[n] == 0).nonzero(as_tuple=False)[:, 0])
         for n in range(N)
     ]

pytorch3d/ops/graph_conv.py

@@ -119,7 +119,6 @@ def gather_scatter_python(input, edges, directed: bool = False):
     idx0 = edges[:, 0].view(num_edges, 1).expand(num_edges, input_feature_dim)
     idx1 = edges[:, 1].view(num_edges, 1).expand(num_edges, input_feature_dim)
 
-    # pyre-fixme[16]: `Tensor` has no attribute `scatter_add`.
     output = output.scatter_add(0, idx0, input.gather(0, idx1))
     if not directed:
         output = output.scatter_add(0, idx1, input.gather(0, idx0))

pytorch3d/ops/interp_face_attrs.py

@@ -94,7 +94,6 @@ def interpolate_face_attributes_python(
     pix_to_face = pix_to_face.clone()
     pix_to_face[mask] = 0
     idx = pix_to_face.view(N * H * W * K, 1, 1).expand(N * H * W * K, 3, D)
-    # pyre-fixme[16]: `Tensor` has no attribute `gather`.
     pixel_face_vals = face_attributes.gather(0, idx).view(N, H, W, K, 3, D)
     pixel_vals = (barycentric_coords[..., None] * pixel_face_vals).sum(dim=-2)
     pixel_vals[mask] = 0  # Replace masked values in output.

pytorch3d/ops/iou_box3d.py

@@ -47,7 +47,6 @@ _box_triangles = [
 
 def _check_coplanar(boxes: torch.Tensor, eps: float = 1e-4) -> None:
     faces = torch.tensor(_box_planes, dtype=torch.int64, device=boxes.device)
-    # pyre-fixme[16]: `boxes` has no attribute `index_select`.
     verts = boxes.index_select(index=faces.view(-1), dim=1)
     B = boxes.shape[0]
     P, V = faces.shape
@@ -74,7 +73,6 @@ def _check_nonzero(boxes: torch.Tensor, eps: float = 1e-4) -> None:
     Checks that the sides of the box have a non zero area
     """
     faces = torch.tensor(_box_triangles, dtype=torch.int64, device=boxes.device)
-    # pyre-fixme[16]: `boxes` has no attribute `index_select`.
     verts = boxes.index_select(index=faces.view(-1), dim=1)
     B = boxes.shape[0]
     T, V = faces.shape

pytorch3d/ops/knn.py

@@ -84,7 +84,6 @@ class _knn_points(Function):
                 dists[mask] = 0
             else:
                 dists, sort_idx = dists.sort(dim=2)
-            # pyre-fixme[16]: `Tensor` has no attribute `gather`.
             idx = idx.gather(2, sort_idx)
 
         ctx.save_for_backward(p1, p2, lengths1, lengths2, idx)

pytorch3d/ops/laplacian_matrices.py

@@ -45,6 +45,7 @@ def laplacian(verts: torch.Tensor, edges: torch.Tensor) -> torch.Tensor:
     # i.e. A[i, j] = 1 if (i,j) is an edge, or
     # A[e0, e1] = 1 &  A[e1, e0] = 1
     ones = torch.ones(idx.shape[1], dtype=torch.float32, device=verts.device)
+    # pyre-fixme[16]: Module `sparse` has no attribute `FloatTensor`.
     A = torch.sparse.FloatTensor(idx, ones, (V, V))
 
     # the sum of i-th row of A gives the degree of the i-th vertex
@@ -53,16 +54,20 @@ def laplacian(verts: torch.Tensor, edges: torch.Tensor) -> torch.Tensor:
     # We construct the Laplacian matrix by adding the non diagonal values
     # i.e. L[i, j] = 1 ./ deg(i) if (i, j) is an edge
     deg0 = deg[e0]
+    # pyre-fixme[58]: `/` is not supported for operand types `float` and `Tensor`.
     deg0 = torch.where(deg0 > 0.0, 1.0 / deg0, deg0)
     deg1 = deg[e1]
+    # pyre-fixme[58]: `/` is not supported for operand types `float` and `Tensor`.
     deg1 = torch.where(deg1 > 0.0, 1.0 / deg1, deg1)
     val = torch.cat([deg0, deg1])
+    # pyre-fixme[16]: Module `sparse` has no attribute `FloatTensor`.
     L = torch.sparse.FloatTensor(idx, val, (V, V))
 
     # Then we add the diagonal values L[i, i] = -1.
     idx = torch.arange(V, device=verts.device)
     idx = torch.stack([idx, idx], dim=0)
     ones = torch.ones(idx.shape[1], dtype=torch.float32, device=verts.device)
+    # pyre-fixme[16]: Module `sparse` has no attribute `FloatTensor`.
     L -= torch.sparse.FloatTensor(idx, ones, (V, V))
 
     return L
@@ -119,6 +124,7 @@ def cot_laplacian(
     ii = faces[:, [1, 2, 0]]
     jj = faces[:, [2, 0, 1]]
     idx = torch.stack([ii, jj], dim=0).view(2, F * 3)
+    # pyre-fixme[16]: Module `sparse` has no attribute `FloatTensor`.
     L = torch.sparse.FloatTensor(idx, cot.view(-1), (V, V))
 
     # Make it symmetric; this means we are also setting
@@ -133,6 +139,7 @@ def cot_laplacian(
     val = torch.stack([area] * 3, dim=1).view(-1)
     inv_areas.scatter_add_(0, idx, val)
     idx = inv_areas > 0
+    # pyre-fixme[58]: `/` is not supported for operand types `float` and `Tensor`.
     inv_areas[idx] = 1.0 / inv_areas[idx]
     inv_areas = inv_areas.view(-1, 1)
 
@@ -166,6 +173,7 @@ def norm_laplacian(
     e01 = edges.t()  # (2, E)
 
     V = verts.shape[0]
+    # pyre-fixme[16]: Module `sparse` has no attribute `FloatTensor`.
     L = torch.sparse.FloatTensor(e01, w01, (V, V))
     L = L + L.t()
 

pytorch3d/ops/marching_cubes.py

@@ -347,4 +347,5 @@ def _get_value(point: Tuple[int, int, int], volume_data: torch.Tensor) -> float:
         data: scalar value in the volume at the given point
     """
     x, y, z = point
+    # pyre-fixme[7]: Expected `float` but got `Tensor`.
     return volume_data[z][y][x]

pytorch3d/ops/mesh_filtering.py

@@ -49,7 +49,6 @@ def taubin_smoothing(
         total_weight = torch.sparse.sum(L, dim=1).to_dense().view(-1, 1)
         verts = (1 - lambd) * verts + lambd * torch.mm(L, verts) / total_weight
 
-        # pyre-ignore
         L = norm_laplacian(verts, edges)
         total_weight = torch.sparse.sum(L, dim=1).to_dense().view(-1, 1)
         verts = (1 - mu) * verts + mu * torch.mm(L, verts) / total_weight

pytorch3d/ops/points_alignment.py

@@ -180,6 +180,7 @@ def iterative_closest_point(
         t_history.append(SimilarityTransform(R, T, s))
 
         # compute the root mean squared error
+        # pyre-fixme[58]: `**` is not supported for operand types `Tensor` and `int`.
         Xt_sq_diff = ((Xt - Xt_nn_points) ** 2).sum(2)
         rmse = oputil.wmean(Xt_sq_diff[:, :, None], mask_X).sqrt()[:, 0, 0]