Fix table formatting in logs; Add support for dicts in PrintData

examples/FasterRCNN/data.py

@@ -22,7 +22,7 @@ from common import (
     filter_boxes_inside_shape, np_iou, point8_to_box, polygons_to_mask,
 )
 from config import config as cfg
-from dataset import DatasetRegistry
+from dataset import DatasetRegistry, register_coco
 from utils.np_box_ops import area as np_area
 from utils.np_box_ops import ioa as np_ioa
 
@@ -50,7 +50,7 @@ def print_class_histogram(roidbs):
         gt_classes = entry["class"][gt_inds]
         gt_hist += np.histogram(gt_classes, bins=hist_bins)[0]
     data = list(itertools.chain(*[[class_names[i + 1], v] for i, v in enumerate(gt_hist[1:])]))
-    COL = max(6, len(data))
+    COL = min(6, len(data))
     total_instances = sum(data[1::2])
     data.extend([None] * (COL - len(data) % COL))
     data.extend(["total", total_instances])
@@ -394,11 +394,12 @@ def get_eval_dataflow(name, shard=0, num_shards=1):
 if __name__ == "__main__":
     import os
     from tensorpack.dataflow import PrintData
+    from config import finalize_configs
 
-    cfg.DATA.BASEDIR = os.path.expanduser("~/data/coco")
+    register_coco(os.path.expanduser("~/data/coco"))
+    finalize_configs()
     ds = get_train_dataflow()
-    ds = PrintData(ds, 100)
+    ds = PrintData(ds, 10)
     TestDataSpeed(ds, 50000).start()
-    ds.reset_state()
     for k in ds:
         pass

tensorpack/dataflow/common.py

@@ -5,7 +5,7 @@ from __future__ import division
 import itertools
 import numpy as np
 import pprint
-from collections import defaultdict, deque
+from collections import defaultdict, deque, Mapping
 from copy import copy
 import six
 import tqdm
@@ -748,7 +748,7 @@ class PrintData(ProxyDataFlow):
         Gather useful debug information from a datapoint.
 
         Args:
-            entry: the datapoint component
+            entry: the datapoint component, either a list or a dict
             k (int): index of this component in current datapoint
             depth (int, optional): recursion depth
             max_depth, max_list: same as in :meth:`__init__`.
@@ -779,7 +779,7 @@ class PrintData(ProxyDataFlow):
                     self.range = " in range [{}, {}]".format(el.min(), el.max())
                 elif type(el) in numpy_scalar_types:
                     self.range = " with value {}".format(el)
-                elif isinstance(el, (list)):
+                elif isinstance(el, (list, tuple)):
                     self.shape = " of len {}".format(len(el))
 
                     if depth < max_depth:
@@ -805,9 +805,15 @@ class PrintData(ProxyDataFlow):
         return str(_elementInfo(entry, k, depth, max_list))
 
     def _get_msg(self, dp):
-        msg = [u"datapoint %i<%i with %i components consists of" % (self.cnt, self.num, len(dp))]
+        msg = [colored(u"datapoint %i/%i with %i components consists of" %
+               (self.cnt, self.num, len(dp)), "cyan")]
+        is_dict = isinstance(dp, Mapping)
         for k, entry in enumerate(dp):
-            msg.append(self._analyze_input_data(entry, k, max_depth=self.max_depth, max_list=self.max_list))
+            if is_dict:
+                key, value = entry, dp[entry]
+            else:
+                key, value = k, entry
+            msg.append(self._analyze_input_data(value, key, max_depth=self.max_depth, max_list=self.max_list))
         return u'\n'.join(msg)
 
     def __iter__(self):
@@ -815,7 +821,7 @@ class PrintData(ProxyDataFlow):
             # it is important to place this here! otherwise it mixes the output of multiple PrintData
             if self.cnt == 0:
                 label = ' (%s)' % self.name if self.name is not None else ""
-                logger.info(colored("DataFlow Info%s:" % label, 'cyan'))
+                logger.info(colored("Contents of DataFlow%s:" % label, 'cyan'))
 
             if self.cnt < self.num:
                 print(self._get_msg(dp))