def selective_search_roidb(self):
"""
Return the database of selective search regions of interest.
Ground-truth ROIs are also included.
This function loads/saves from/to a cache file to speed up future calls.
"""
cache_file = os.path.join(self.cache_path,
self.name + '_selective_search_roidb.pkl')
if os.path.exists(cache_file):
with open(cache_file, 'rb') as fid:
roidb = pickle.load(fid)
print('{} ss roidb loaded from {}'.format(self.name, cache_file))
return roidb
if int(self._year) == 2007 or self._image_set != 'test':
gt_roidb = self.gt_roidb()
ss_roidb = self._load_selective_search_roidb(gt_roidb)
roidb = imdb.merge_roidbs(gt_roidb, ss_roidb)
else:
roidb = self._load_selective_search_roidb(None)
with open(cache_file, 'wb') as fid:
pickle.dump(roidb, fid, pickle.HIGHEST_PROTOCOL)
print('wrote ss roidb to {}'.format(cache_file))
return roidb
def _roidb_from_proposals(self, method):
"""
Creates a roidb from pre-computed proposals of a particular methods.
"""
top_k = self.config['top_k']
cache_file = osp.join(
self.cache_path,
self.name + '_{:s}_top{:d}'.format(method, top_k) + '_roidb.pkl')
if osp.exists(cache_file):
with open(cache_file, 'rb') as fid:
roidb = cPickle.load(fid)
print '{:s} {:s} roidb loaded from {:s}'.format(
self.name, method, cache_file)
return roidb
if self._image_set in self._gt_splits:
gt_roidb = self.gt_roidb()
method_roidb = self._load_proposals(method, gt_roidb)
roidb = imdb.merge_roidbs(gt_roidb, method_roidb)
# Make sure we don't use proposals that are contained in crowds
roidb = _filter_crowd_proposals(roidb, self.config['crowd_thresh'])
else:
roidb = self._load_proposals(method, None)
with open(cache_file, 'wb') as fid:
cPickle.dump(roidb, fid, cPickle.HIGHEST_PROTOCOL)
print 'wrote {:s} roidb to {:s}'.format(method, cache_file)
return roidb
def rpn_roidb(self):
if self._image_set != 'test':
gt_roidb = self.gt_roidb()
rpn_roidb = self._load_rpn_roidb(gt_roidb)
roidb = imdb.merge_roidbs(gt_roidb, rpn_roidb)
else:
roidb = self._load_rpn_roidb(None)
return roidb
def rpn_roidb(self):
if int(self._year) == 2007 or self._image_set != 'valid':
gt_roidb = self.gt_roidb()
rpn_roidb = self._load_rpn_roidb(gt_roidb)
roidb = imdb.merge_roidbs(gt_roidb, rpn_roidb)
else:
roidb = self._load_rpn_roidb(None)
return roidb
def rpn_roidb(self):
# 在经过RPN网络产生了proposal以后，这个函数作用是将这些proposal
# 的
# roi与groudtruth结合起来，送入网络训练。
# 那怎么个结合法呢？proposal
# 的
# roidb格式与上面提到的gt_roidb一模一样，只不过overlap由1变成了与最接近的class的重合度。
# 如何判断是最接近的class呢？每个proposal的box都与groud - truth的box做一次重合度计算，与anchor
# _target
# _layer.py中类似
# overlap = (重合部分面积) / (proposal _box面积 + gt_boxes面积 - 重合部分面积)
# 对于每个proposal，选出最大的那个gt
# _boxes的值，然后填到相应的class
# index下。
if int(self._year) == 2007 or self._image_set != 'test':
gt_roidb = self.gt_roidb()
rpn_roidb = self._load_rpn_roidb(gt_roidb)
roidb = imdb.merge_roidbs(gt_roidb, rpn_roidb)
else:
roidb = self._load_rpn_roidb(None)
return roidb
