[MaskRCNN] bugfix on data loading

examples/FasterRCNN/coco.py

@@ -144,7 +144,7 @@ class COCODetection(object):
                         assert obj['iscrowd'] == 1
                         obj['segmentation'] = None
                     else:
-                        valid_segs = [np.asarray(p).reshape(-1, 2) for p in segs if len(p) >= 6]
+                        valid_segs = [np.asarray(p).reshape(-1, 2).astype('float32') for p in segs if len(p) >= 6]
                         if len(valid_segs) < len(segs):
                             log_once("Image {} has invalid polygons!".format(img['file_name']), 'warn')
 
@@ -164,7 +164,7 @@ class COCODetection(object):
         if add_mask:
             # also required to be float32
             img['segmentation'] = [
-                obj['segmentation'].astype('float32') for obj in valid_objs]
+                obj['segmentation'] for obj in valid_objs]
 
     def print_class_histogram(self, imgs):
         nr_class = len(COCOMeta.class_names)

examples/FasterRCNN/data.py

@@ -272,8 +272,12 @@ def get_train_dataflow():
     boxes: kx4 floats
     class: k integers
     is_crowd: k booleans. Use k False if you don't know what it means.
-    segmentation: k numpy arrays. Each array is a polygon of shape Nx2.
-        If your segmentation annotations are masks rather than polygons,
+    segmentation: k lists of numpy arrays (one for each box).
+        Each list of numpy array corresponds to the mask for one instance.
+        Each numpy array in the list is a polygon of shape Nx2,
+        because one mask can be represented by N polygons.
+
+        If your segmentation annotations are originally masks rather than polygons,
         either convert it, or the augmentation code below will need to be
         changed or skipped accordingly.
     """
@@ -369,7 +373,6 @@ def get_eval_dataflow():
 
 if __name__ == '__main__':
     import os
-    # import IPython as IP; IP.embed()
     from tensorpack.dataflow import PrintData
     config.BASEDIR = os.path.expanduser('~/data/coco')
     ds = get_train_dataflow()

examples/FasterRCNN/model.py

@@ -377,7 +377,7 @@ def crop_and_resize(image, boxes, box_ind, crop_size, pad_border=True):
     boxes = transform_fpcoor_for_tf(boxes, image_shape, [crop_size, crop_size])
     image = tf.transpose(image, [0, 2, 3, 1])   # 1hwc
     ret = tf.image.crop_and_resize(
-        image, boxes, box_ind,
+        image, boxes, tf.to_int32(box_ind),
         crop_size=[crop_size, crop_size])
     ret = tf.transpose(ret, [0, 3, 1, 2])   # ncss
     return ret