Merge pull request #830 from 23pointsNorth/fpn_param

FPN - Add min/max size params for pre-processing.

chainercv/links/model/fpn/faster_rcnn.py

@@ -26,6 +26,11 @@ class FasterRCNN(chainer.Chain):
         head (Link): A link that has the same interface as
             :class:`~chainercv.links.model.fpn.Head`.
             Please refer to the documentation found there.
+        min_size (int): A preprocessing paramter for :meth:`prepare`. Please
+            refer to a docstring found for :meth:`prepare`.
+        max_size (int): A preprocessing paramter for :meth:`prepare`. Note
+            that the result of :meth:`prepare` can exceed this size due to
+            alignment with stride.
 
     Parameters:
         nms_thresh (float): The threshold value
@@ -40,17 +45,19 @@ class FasterRCNN(chainer.Chain):
 
     """
 
-    _min_size = 800
-    _max_size = 1333
     _stride = 32
 
-    def __init__(self, extractor, rpn, head):
+    def __init__(self, extractor, rpn, head,
+                 min_size=800, max_size=1333):
         super(FasterRCNN, self).__init__()
         with self.init_scope():
             self.extractor = extractor
             self.rpn = rpn
             self.head = head
 
+        self._min_size = min_size
+        self._max_size = max_size
+
         self.use_preset('visualize')
 
     def use_preset(self, preset):

chainercv/links/model/fpn/faster_rcnn_fpn_resnet.py

@@ -19,7 +19,8 @@ class FasterRCNNFPNResNet(FasterRCNN):
     A subclass of this class should have :obj:`_base` and :obj:`_models`.
     """
 
-    def __init__(self, n_fg_class=None, pretrained_model=None):
+    def __init__(self, n_fg_class=None, pretrained_model=None,
+                 min_size=800, max_size=1333):
         param, path = utils.prepare_pretrained_model(
             {'n_fg_class': n_fg_class}, pretrained_model, self._models)
 
@@ -35,6 +36,7 @@ def __init__(self, n_fg_class=None, pretrained_model=None):
             extractor=extractor,
             rpn=RPN(extractor.scales),
             head=Head(param['n_fg_class'] + 1, extractor.scales),
+            min_size=min_size, max_size=max_size
         )
 
         if path == 'imagenet':
@@ -56,10 +58,10 @@ class FasterRCNNFPNResNet50(FasterRCNNFPNResNet):
        Feature Pyramid Networks for Object Detection. CVPR 2017
 
     Args:
-       n_fg_class (int): The number of classes excluding the background.
-       pretrained_model (string): The weight file to be loaded.
-           This can take :obj:`'coco'`, `filepath` or :obj:`None`.
-           The default value is :obj:`None`.
+        n_fg_class (int): The number of classes excluding the background.
+        pretrained_model (string): The weight file to be loaded.
+            This can take :obj:`'coco'`, `filepath` or :obj:`None`.
+            The default value is :obj:`None`.
 
             * :obj:`'coco'`: Load weights trained on train split of \
                 MS COCO 2017. \
@@ -74,6 +76,9 @@ class FasterRCNNFPNResNet50(FasterRCNNFPNResNet):
             * `filepath`: A path of npz file. In this case, :obj:`n_fg_class` \
                 must be specified properly.
             * :obj:`None`: Do not load weights.
+        min_size (int): A preprocessing paramter for :meth:`prepare`. Please \
+            refer to :meth:`prepare`.
+        max_size (int): A preprocessing paramter for :meth:`prepare`.
 
     """
 
@@ -99,10 +104,10 @@ class FasterRCNNFPNResNet101(FasterRCNNFPNResNet):
        Feature Pyramid Networks for Object Detection. CVPR 2017
 
     Args:
-       n_fg_class (int): The number of classes excluding the background.
-       pretrained_model (string): The weight file to be loaded.
-           This can take :obj:`'coco'`, `filepath` or :obj:`None`.
-           The default value is :obj:`None`.
+        n_fg_class (int): The number of classes excluding the background.
+        pretrained_model (string): The weight file to be loaded.
+            This can take :obj:`'coco'`, `filepath` or :obj:`None`.
+            The default value is :obj:`None`.
 
             * :obj:`'coco'`: Load weights trained on train split of \
                 MS COCO 2017. \
@@ -117,6 +122,9 @@ class FasterRCNNFPNResNet101(FasterRCNNFPNResNet):
             * `filepath`: A path of npz file. In this case, :obj:`n_fg_class` \
                 must be specified properly.
             * :obj:`None`: Do not load weights.
+        min_size (int): A preprocessing paramter for :meth:`prepare`. Please \
+            refer to :meth:`prepare`.
+        max_size (int): A preprocessing paramter for :meth:`prepare`.
 
     """
 

tests/links_tests/model_tests/fpn_tests/test_faster_rcnn.py

@@ -31,24 +31,41 @@ def __call__(self, x):
 
 class DummyFasterRCNN(FasterRCNN):
 
-    def __init__(self, n_fg_class):
+    def __init__(self, n_fg_class, min_size, max_size):
         extractor = DummyExtractor()
         super(DummyFasterRCNN, self).__init__(
             extractor=extractor,
             rpn=RPN(extractor.scales),
             head=Head(n_fg_class + 1, extractor.scales),
+            min_size=min_size, max_size=max_size,
         )
 
 
-@testing.parameterize(
-    {'n_fg_class': 1},
-    {'n_fg_class': 5},
-    {'n_fg_class': 20},
-)
+@testing.parameterize(*testing.product_dict(
+    [
+        {'n_fg_class': 1},
+        {'n_fg_class': 5},
+        {'n_fg_class': 20},
+    ],
+    [
+        {
+            'in_sizes': [(480, 640), (320, 320)],
+            'min_size': 800, 'max_size': 1333,
+            'expected_shape': (800, 1088),
+        },
+        {
+            'in_sizes': [(200, 50), (400, 100)],
+            'min_size': 200, 'max_size': 320,
+            'expected_shape': (320, 96),
+        },
+    ],
+))
 class TestFasterRCNN(unittest.TestCase):
 
     def setUp(self):
-        self.link = DummyFasterRCNN(n_fg_class=self.n_fg_class)
+        self.link = DummyFasterRCNN(n_fg_class=self.n_fg_class,
+                                    min_size=self.min_size,
+                                    max_size=self.max_size)
 
     def test_use_preset(self):
         self.link.nms_thresh = 0
@@ -118,12 +135,13 @@ def test_predict_gpu(self):
         assert_is_detection_link(self.link, self.n_fg_class)
 
     def test_prepare(self):
-        imgs = [
-            np.random.randint(0, 255, size=(3, 480, 640)).astype(np.float32),
-            np.random.randint(0, 255, size=(3, 320, 320)).astype(np.float32),
-        ]
-        x, scales = self.link.prepare(imgs)
-        self.assertEqual(x.shape, (2, 3, 800, 1088))
+        imgs = [_random_array(np, (3, s[0], s[1])) for s in self.in_sizes]
+        out, scales = self.link.prepare(imgs)
+        self.assertIsInstance(out, np.ndarray)
+        full_expected_shape = (len(self.in_sizes), 3,
+                               self.expected_shape[0],
+                               self.expected_shape[1])
+        self.assertEqual(out.shape, full_expected_shape)
 
 
 testing.run_module(__name__, __file__)