add cascade rcnn model

MODEL_ZOO.md

@@ -79,6 +79,14 @@ We released RPN, Faster R-CNN and Mask R-CNN models in the first version. More m
 | R-50-FPN | pytorch | 1x      |          |                     |                |        |          |
 | R-50-FPN | pytorch | 2x      |          |                     |                |        |          |
 
+### Cascade R-CNN
+
+| Backbone | Style   | Lr schd | Mem (GB) | Train time (s/iter) | Inf time (fps) | box AP | Download |
+|:--------:|:-------:|:-------:|:--------:|:-------------------:|:--------------:|:------:|:--------:|
+| R-50-FPN | caffe   | 1x      | 5.0      | 0.592               | 8.1            | 40.3   | -        |
+| R-50-FPN | pytorch | 1x      | 5.5      | 0.622               | 8.0            | 40.3   | [model](https://s3.ap-northeast-2.amazonaws.com/open-mmlab/mmdetection/models/cascade_rcnn_r50_fpn_1x_20181123-c019abac.pth) |
+| R-50-FPN | pytorch | 20e     | 5.5      | 0.622               | 8.0            | 41.1   | [model](https://s3.ap-northeast-2.amazonaws.com/open-mmlab/mmdetection/models/cascade_rcnn_r50_fpn_20e_20181123-c4cad178.pth) |
+
 
 ## Comparison with Detectron
 

mmdet/models/bbox_heads/bbox_head.py

@@ -171,10 +171,10 @@ class BBoxHead(nn.Module):
             rois (Tensor): shape (n, 4) or (n, 5)
             label (Tensor): shape (n, )
             bbox_pred (Tensor): shape (n, 4*(#class+1)) or (n, 4)
-            img_shape (Tensor): shape (3, )
+            img_meta (dict): Image meta info.
 
         Returns:
-            Tensor: same shape as input rois
+            Tensor: Regressed bboxes, the same shape as input rois.
         """
         assert rois.size(1) == 4 or rois.size(1) == 5