changed according to

matterport#1070 (comment)

mrcnn/model.py

@@ -699,7 +699,7 @@ def refine_detections_graph(rois, probs, deltas, window, config):
     # Class IDs per ROI
     class_ids = tf.argmax(probs, axis=1, output_type=tf.int32)
     # Class probability of the top class of each ROI
-    indices = tf.stack([tf.range(probs.shape[0]), class_ids], axis=1)
+    indices = tf.stack([tf.range(tf.shape(probs)[0]), class_ids], axis = 1)
     class_scores = tf.gather_nd(probs, indices)
     # Class-specific bounding box deltas
     deltas_specific = tf.gather_nd(deltas, indices)
@@ -948,7 +948,10 @@ def fpn_classifier_graph(rois, feature_maps, image_meta,
                            name='mrcnn_bbox_fc')(shared)
     # Reshape to [batch, num_rois, NUM_CLASSES, (dy, dx, log(dh), log(dw))]
     s = K.int_shape(x)
-    mrcnn_bbox = KL.Reshape((s[1], num_classes, 4), name="mrcnn_bbox")(x)
+    if s[1]==None:
+	mrcnn_bbox = layers.Reshape((-1, num_classes, 4), name="mrcnn_bbox")(x)
+	else:
+		mrcnn_bbox = layers.Reshape((s[1], num_classes, 4), name="mrcnn_bbox")(x)
 
     return mrcnn_class_logits, mrcnn_probs, mrcnn_bbox