[FasterRCNN] update docs; fix tf.reverse

examples/FasterRCNN/NOTES.md

@@ -5,7 +5,7 @@ This is a minimal implementation that simply contains these files:
 + data.py: prepare data for training
 + common.py: common data preparation utilities
 + basemodel.py: implement resnet
-+ model.py: implement rpn/faster-rcnn
++ model.py: implement rpn/faster-rcnn/mask-rcnn
 + train.py: main training script
 + utils/: third-party helper functions
 + eval.py: evaluation utilities
@@ -45,5 +45,5 @@ Speed:
 	 a slow convolution algorithm, or you spend more time on autotune.
 	 This is a general problem of TensorFlow when running against variable-sized input.
 
-3. With a large roi batch size (e.g. >= 256), GPU utilitization should stay around 90%.
+3. With a large roi batch size (e.g. >= 256), GPU utilitization should stay above 90%.
 

examples/FasterRCNN/README.md

@@ -24,8 +24,8 @@ DIR/
 
 
 ## Usage
-Change config:
-1. Set `BASEDIR` in `config.py` to `/path/to/DIR` as described above.
+Change config in `config.py`:
+1. Set `BASEDIR` to `/path/to/DIR` as described above.
 2. Set `MODE_MASK` to switch Faster-RCNN or Mask-RCNN.
 
 Train:

examples/FasterRCNN/train.py

@@ -147,7 +147,7 @@ class Model(ModelDesc):
                     image, fg_sampled_boxes,
                     tf.zeros_like(fg_inds_wrt_sample, dtype=tf.int32), 300)
                 fg_sampled_patches = tf.transpose(fg_sampled_patches, [0, 2, 3, 1])
-                fg_sampled_patches = tf.reverse(fg_sampled_patches, axis=-1)  # BGR->RGB
+                fg_sampled_patches = tf.reverse(fg_sampled_patches, axis=[-1])  # BGR->RGB
                 tf.summary.image('viz', fg_sampled_patches, max_outputs=30)
 
             matched_gt_boxes = tf.gather(gt_boxes, fg_inds_wrt_gt)

tensorpack/tfutils/varmanip.py

@@ -169,7 +169,7 @@ def get_checkpoint_path(model_path):
 
 
 def load_chkpt_vars(model_path):
-    """ Dump all variables from a checkpoint to a dict.
+    """ Load all variables from a checkpoint to a dict.
 
     Args:
         model_path(str): path to a checkpoint.