FPN initial commit

examples/FasterRCNN/basemodel.py

@@ -117,9 +117,18 @@ def resnet_conv5(image, num_block):
 
 
 def pretrained_resnet_fpn_backbone(image, num_blocks, freeze_c2=True):
+    shape2d = tf.shape(image)[2:]
+    mult = config.FPN_RESOLUTION_REQUIREMENT * 1.
+    new_shape2d = tf.to_int32(tf.ceil(tf.to_float(shape2d) / mult) * mult)
+    pad_shape2d = new_shape2d - shape2d
     assert len(num_blocks) == 4
+    # TODO pad 1 at each stage
     with resnet_argscope():
-        l = tf.pad(image, [[0, 0], [0, 0], [2, 3], [2, 3]])
+        chan = image.shape[1]
+        l = tf.pad(image,
+                   tf.stack([[0, 0], [0, 0],
+                            [2, 3 + pad_shape2d[0]], [2, 3 + pad_shape2d[1]]]))
+        l.set_shape([None, chan, None, None])
         l = Conv2D('conv0', l, 64, 7, strides=2, activation=BNReLU, padding='VALID')
         l = tf.pad(l, [[0, 0], [0, 0], [0, 1], [0, 1]])
         l = MaxPooling('pool0', l, 3, strides=2, padding='VALID')

examples/FasterRCNN/config.py

@@ -4,11 +4,12 @@
 import numpy as np
 
 # mode flags ---------------------
-MODE_MASK = True
+MODE_MASK = False
 
 # dataset -----------------------
 BASEDIR = '/path/to/your/COCO/DIR'
 TRAIN_DATASET = ['train2014', 'valminusminival2014']
+# TRAIN_DATASET = ['valminusminival2014']
 VAL_DATASET = 'minival2014'   # only support evaluation on single dataset
 NUM_CLASS = 81
 CLASS_NAMES = []  # NUM_CLASS strings. Will be populated later by coco loader
@@ -23,12 +24,12 @@ BASE_LR = 1e-2
 WARMUP = 1000    # in steps
 STEPS_PER_EPOCH = 500
 LR_SCHEDULE = [150000, 230000, 280000]
-# LR_SCHEDULE = [120000, 160000, 180000]    # "1x" schedule in detectron
+LR_SCHEDULE = [120000, 160000, 180000]    # "1x" schedule in detectron
 # LR_SCHEDULE = [240000, 320000, 360000]    # "2x" schedule in detectron
 
 # image resolution --------------------
 SHORT_EDGE_SIZE = 800
-MAX_SIZE = 1333
+MAX_SIZE = 1333  # TODO use 1344
 # alternative (worse & faster) setting: 600, 1024
 
 # anchors -------------------------
@@ -55,7 +56,7 @@ TRAIN_POST_NMS_TOPK = 2000
 CROWD_OVERLAP_THRES = 0.7
 
 # fastrcnn training ---------------------
-FASTRCNN_BATCH_PER_IM = 256
+FASTRCNN_BATCH_PER_IM = 512
 FASTRCNN_BBOX_REG_WEIGHTS = np.array([10, 10, 5, 5], dtype='float32')
 FASTRCNN_FG_THRESH = 0.5
 # fg ratio in a ROI batch
@@ -70,6 +71,9 @@ RESULT_SCORE_THRESH_VIS = 0.3   # only visualize confident results
 RESULTS_PER_IM = 100
 
 # TODO Not Functioning. Don't USE
-MODE_FPN = False
+MODE_FPN = True
 FPN_NUM_CHANNEL = 256
-FPN_SIZE_REQUIREMENT = 32
+FASTRCNN_FC_HEAD_DIM = 1024
+FPN_RESOLUTION_REQUIREMENT = 32
+TRAIN_FPN_NMS_TOPK = 2048
+TEST_FPN_NMS_TOPK = 1024

examples/FasterRCNN/data.py

@@ -51,7 +51,12 @@ def get_all_anchors(
     # anchors are intbox here.
     # anchors at featuremap [0,0] are centered at fpcoor (8,8) (half of stride)
 
-    field_size = int(np.ceil(config.MAX_SIZE / stride))
+    max_size = config.MAX_SIZE
+    if config.MODE_FPN:
+        # TODO setting this in config is perhaps better
+        size_mult = config.FPN_RESOLUTION_REQUIREMENT * 1.
+        max_size = np.ceil(max_size / size_mult) * size_mult
+    field_size = int(np.ceil(max_size / stride))
     shifts = np.arange(0, field_size) * stride
     shift_x, shift_y = np.meshgrid(shifts, shifts)
     shift_x = shift_x.flatten()
@@ -337,7 +342,7 @@ def get_train_dataflow(add_mask=False):
         return ret
 
     ds = MapData(ds, preprocess)
-    ds = PrefetchDataZMQ(ds, 1)
+    ds = PrefetchDataZMQ(ds, 3)
     return ds
 
 
@@ -359,7 +364,6 @@ if __name__ == '__main__':
     import os
     from tensorpack.dataflow import PrintData
     config.BASEDIR = os.path.expanduser('~/data/coco')
-    config.TRAIN_DATASET = ['train2014']
     ds = get_train_dataflow(add_mask=config.MODE_MASK)
     ds = PrintData(ds, 100)
     TestDataSpeed(ds, 50000).start()