Rename mask field to masks (#934)

This makes it consistent with the other models, which returns nouns in plurial

references/detection/coco_eval.py

@@ -101,7 +101,7 @@ class CocoEvaluator(object):
 
             scores = prediction["scores"]
             labels = prediction["labels"]
-            masks = prediction["mask"]
+            masks = prediction["masks"]
 
             masks = masks > 0.5
 

torchvision/models/detection/mask_rcnn.py

@@ -43,7 +43,7 @@ class MaskRCNN(FasterRCNN):
           0 and H and 0 and W
         - labels (Tensor[N]): the predicted labels for each image
         - scores (Tensor[N]): the scores or each prediction
-        - mask (Tensor[N, H, W]): the predicted masks for each instance, in 0-1 range. In order to
+        - masks (Tensor[N, H, W]): the predicted masks for each instance, in 0-1 range. In order to
           obtain the final segmentation masks, the soft masks can be thresholded, generally
           with a value of 0.5 (mask >= 0.5)
 
@@ -249,7 +249,7 @@ def maskrcnn_resnet50_fpn(pretrained=False, progress=True,
           ``0`` and ``H`` and ``0`` and ``W``
         - labels (``Tensor[N]``): the predicted labels for each image
         - scores (``Tensor[N]``): the scores or each prediction
-        - mask (``Tensor[N, H, W]``): the predicted masks for each instance, in ``0-1`` range. In order to
+        - masks (``Tensor[N, H, W]``): the predicted masks for each instance, in ``0-1`` range. In order to
           obtain the final segmentation masks, the soft masks can be thresholded, generally
           with a value of 0.5 (``mask >= 0.5``)
 

torchvision/models/detection/roi_heads.py

@@ -573,7 +573,7 @@ class RoIHeads(torch.nn.Module):
                 labels = [r["labels"] for r in result]
                 masks_probs = maskrcnn_inference(mask_logits, labels)
                 for mask_prob, r in zip(masks_probs, result):
-                    r["mask"] = mask_prob
+                    r["masks"] = mask_prob
 
             losses.update(loss_mask)
 

torchvision/models/detection/transform.py

@@ -109,10 +109,10 @@ class GeneralizedRCNNTransform(nn.Module):
             boxes = pred["boxes"]
             boxes = resize_boxes(boxes, im_s, o_im_s)
             result[i]["boxes"] = boxes
-            if "mask" in pred:
-                masks = pred["mask"]
+            if "masks" in pred:
+                masks = pred["masks"]
                 masks = paste_masks_in_image(masks, boxes, o_im_s)
-                result[i]["mask"] = masks
+                result[i]["masks"] = masks
             if "keypoints" in pred:
                 keypoints = pred["keypoints"]
                 keypoints = resize_keypoints(keypoints, im_s, o_im_s)