Adding support to Caffe2 ResNeXt-152-32x8d-FPN-IN5k backbone for Mask R-CNN (#405)

configs/caffe2/e2e_mask_rcnn_X-152-32x8d-FPN-IN5k_1.44x_caffe2.yaml

+MODEL:
+  META_ARCHITECTURE: "GeneralizedRCNN"
+  WEIGHT: "catalog://Caffe2Detectron/COCO/37129812/e2e_mask_rcnn_X-152-32x8d-FPN-IN5k_1.44x"
+  BACKBONE:
+    CONV_BODY: "R-152-FPN"
+    OUT_CHANNELS: 256
+  RPN:
+    USE_FPN: True
+    ANCHOR_STRIDE: (4, 8, 16, 32, 64)
+    PRE_NMS_TOP_N_TRAIN: 2000
+    PRE_NMS_TOP_N_TEST: 1000
+    POST_NMS_TOP_N_TEST: 1000
+    FPN_POST_NMS_TOP_N_TEST: 1000
+  ROI_HEADS:
+    USE_FPN: True
+  ROI_BOX_HEAD:
+    POOLER_RESOLUTION: 7
+    POOLER_SCALES: (0.25, 0.125, 0.0625, 0.03125)
+    POOLER_SAMPLING_RATIO: 2
+    FEATURE_EXTRACTOR: "FPN2MLPFeatureExtractor"
+    PREDICTOR: "FPNPredictor"
+  ROI_MASK_HEAD:
+    POOLER_SCALES: (0.25, 0.125, 0.0625, 0.03125)
+    FEATURE_EXTRACTOR: "MaskRCNNFPNFeatureExtractor"
+    PREDICTOR: "MaskRCNNC4Predictor"
+    POOLER_RESOLUTION: 14
+    POOLER_SAMPLING_RATIO: 2
+    RESOLUTION: 28
+    SHARE_BOX_FEATURE_EXTRACTOR: False
+  RESNETS:
+    STRIDE_IN_1X1: False
+    NUM_GROUPS: 32
+    WIDTH_PER_GROUP: 8
+  MASK_ON: True
+DATASETS:
+  TEST: ("coco_2014_minival",)
+DATALOADER:
+  SIZE_DIVISIBILITY: 32

maskrcnn_benchmark/config/paths_catalog.py

@@ -137,6 +137,7 @@ class ModelCatalog(object):
         "35858933/e2e_mask_rcnn_R-50-FPN_1x": "01_48_14.DzEQe4wC",
         "35861795/e2e_mask_rcnn_R-101-FPN_1x": "02_31_37.KqyEK4tT",
         "36761843/e2e_mask_rcnn_X-101-32x8d-FPN_1x": "06_35_59.RZotkLKI",
+        "37129812/e2e_mask_rcnn_X-152-32x8d-FPN-IN5k_1.44x": "09_35_36.8pzTQKYK",
     }
 
     @staticmethod

maskrcnn_benchmark/modeling/backbone/backbone.py

@@ -21,6 +21,7 @@ def build_resnet_backbone(cfg):
 
 @registry.BACKBONES.register("R-50-FPN")
 @registry.BACKBONES.register("R-101-FPN")
+@registry.BACKBONES.register("R-152-FPN")
 def build_resnet_fpn_backbone(cfg):
     body = resnet.ResNet(cfg)
     in_channels_stage2 = cfg.MODEL.RESNETS.RES2_OUT_CHANNELS

maskrcnn_benchmark/modeling/backbone/resnet.py

@@ -71,7 +71,11 @@ ResNet101FPNStagesTo5 = tuple(
     StageSpec(index=i, block_count=c, return_features=r)
     for (i, c, r) in ((1, 3, True), (2, 4, True), (3, 23, True), (4, 3, True))
 )
-
+# ResNet-152-FPN (including all stages)
+ResNet152FPNStagesTo5 = tuple(
+    StageSpec(index=i, block_count=c, return_features=r)
+    for (i, c, r) in ((1, 3, True), (2, 8, True), (3, 36, True), (4, 3, True))
+)
 
 class ResNet(nn.Module):
     def __init__(self, cfg):
@@ -239,7 +243,7 @@ class Bottleneck(nn.Module):
             down_stride = stride if dilation == 1 else 1
             self.downsample = nn.Sequential(
                 Conv2d(
-                    in_channels, out_channels, 
+                    in_channels, out_channels,
                     kernel_size=1, stride=down_stride, bias=False
                 ),
                 norm_func(out_channels),
@@ -407,4 +411,5 @@ _STAGE_SPECS = Registry({
     "R-101-C5": ResNet101StagesTo5,
     "R-50-FPN": ResNet50FPNStagesTo5,
     "R-101-FPN": ResNet101FPNStagesTo5,
+    "R-152-FPN": ResNet152FPNStagesTo5,
 })

maskrcnn_benchmark/utils/c2_model_loading.py

@@ -146,6 +146,7 @@ def _load_c2_pickled_weights(file_path):
 _C2_STAGE_NAMES = {
     "R-50": ["1.2", "2.3", "3.5", "4.2"],
     "R-101": ["1.2", "2.3", "3.22", "4.2"],
+    "R-152": ["1.2", "2.7", "3.35", "4.2"],
 }
 
 C2_FORMAT_LOADER = Registry()
@@ -157,6 +158,7 @@ C2_FORMAT_LOADER = Registry()
 @C2_FORMAT_LOADER.register("R-101-C5")
 @C2_FORMAT_LOADER.register("R-50-FPN")
 @C2_FORMAT_LOADER.register("R-101-FPN")
+@C2_FORMAT_LOADER.register("R-152-FPN")
 def load_resnet_c2_format(cfg, f):
     state_dict = _load_c2_pickled_weights(f)
     conv_body = cfg.MODEL.BACKBONE.CONV_BODY