Make YOLOv3 neck more flexible (#5218)

* Make YOLOv3 neck more flexible

* Fix linter warning

* update docstring

* add unit test of yolov3 neck

* update

* fix docstring

Co-authored-by: RangiLyu <lyuchqi@gmail.com>

mmdet/models/necks/yolo_neck.py

@@ -76,12 +76,13 @@ class YOLOV3Neck(BaseModule):
 
     Args:
         num_scales (int): The number of scales / stages.
-        in_channels (int): The number of input channels.
-        out_channels (int): The number of output channels.
-        conv_cfg (dict): Config dict for convolution layer. Default: None.
-        norm_cfg (dict): Dictionary to construct and config norm layer.
-            Default: dict(type='BN', requires_grad=True)
-        act_cfg (dict): Config dict for activation layer.
+        in_channels (List[int]): The number of input channels per scale.
+        out_channels (List[int]): The number of output channels  per scale.
+        conv_cfg (dict, optional): Config dict for convolution layer.
+            Default: None.
+        norm_cfg (dict, optional): Dictionary to construct and config norm
+            layer. Default: dict(type='BN', requires_grad=True)
+        act_cfg (dict, optional): Config dict for activation layer.
             Default: dict(type='LeakyReLU', negative_slope=0.1).
         init_cfg (dict or list[dict], optional): Initialization config dict.
             Default: None
@@ -109,7 +110,8 @@ def __init__(self,
         self.detect1 = DetectionBlock(in_channels[0], out_channels[0], **cfg)
         for i in range(1, self.num_scales):
             in_c, out_c = self.in_channels[i], self.out_channels[i]
-            self.add_module(f'conv{i}', ConvModule(in_c, out_c, 1, **cfg))
+            inter_c = out_channels[i - 1]
+            self.add_module(f'conv{i}', ConvModule(inter_c, out_c, 1, **cfg))
             # in_c + out_c : High-lvl feats will be cat with low-lvl feats
             self.add_module(f'detect{i+1}',
                             DetectionBlock(in_c + out_c, out_c, **cfg))

tests/test_models/test_necks.py

@@ -2,7 +2,8 @@
 import torch
 from torch.nn.modules.batchnorm import _BatchNorm
 
-from mmdet.models.necks import FPN, ChannelMapper, CTResNetNeck, DilatedEncoder
+from mmdet.models.necks import (FPN, ChannelMapper, CTResNetNeck,
+                                DilatedEncoder, YOLOV3Neck)
 
 
 def test_fpn():
@@ -288,3 +289,52 @@ def test_ct_resnet_neck():
         feat = feat.cuda()
         out_feat = ct_resnet_neck([feat])[0]
         assert out_feat.shape == (1, num_filters[-1], 16, 16)
+
+
+def test_yolov3_neck():
+    # num_scales, in_channels, out_channels must be same length
+    with pytest.raises(AssertionError):
+        YOLOV3Neck(num_scales=3, in_channels=[16, 8, 4], out_channels=[8, 4])
+
+    # len(feats) must equal to num_scales
+    with pytest.raises(AssertionError):
+        neck = YOLOV3Neck(
+            num_scales=3, in_channels=[16, 8, 4], out_channels=[8, 4, 2])
+        feats = (torch.rand(1, 4, 16, 16), torch.rand(1, 8, 16, 16))
+        neck(feats)
+
+    # test normal channels
+    s = 32
+    in_channels = [16, 8, 4]
+    out_channels = [8, 4, 2]
+    feat_sizes = [s // 2**i for i in range(len(in_channels) - 1, -1, -1)]
+    feats = [
+        torch.rand(1, in_channels[i], feat_sizes[i], feat_sizes[i])
+        for i in range(len(in_channels) - 1, -1, -1)
+    ]
+    neck = YOLOV3Neck(
+        num_scales=3, in_channels=in_channels, out_channels=out_channels)
+    outs = neck(feats)
+
+    assert len(outs) == len(feats)
+    for i in range(len(outs)):
+        assert outs[i].shape == \
+               (1, out_channels[i], feat_sizes[i], feat_sizes[i])
+
+    # test more flexible setting
+    s = 32
+    in_channels = [32, 8, 16]
+    out_channels = [19, 21, 5]
+    feat_sizes = [s // 2**i for i in range(len(in_channels) - 1, -1, -1)]
+    feats = [
+        torch.rand(1, in_channels[i], feat_sizes[i], feat_sizes[i])
+        for i in range(len(in_channels) - 1, -1, -1)
+    ]
+    neck = YOLOV3Neck(
+        num_scales=3, in_channels=in_channels, out_channels=out_channels)
+    outs = neck(feats)
+
+    assert len(outs) == len(feats)
+    for i in range(len(outs)):
+        assert outs[i].shape == \
+               (1, out_channels[i], feat_sizes[i], feat_sizes[i])