def compute_score(self, gts, res):
"""
Main function to compute CIDEr score
:param hypo_for_image (dict) : dictionary with key <image> and value <tokenized hypothesis / candidate sentence>
ref_for_image (dict) : dictionary with key <image> and value <tokenized reference sentence>
:return: cider (float) : computed CIDEr score for the corpus
"""
assert(gts.keys() == res.keys())
imgIds = gts.keys()
cider_scorer = CiderScorer(n=self._n, sigma=self._sigma)
for id in imgIds:
hypo = res[id]
ref = gts[id]
# Sanity check.
assert(type(hypo) is list)
assert(len(hypo) == 1)
assert(type(ref) is list)
assert(len(ref) > 0)
cider_scorer += (hypo[0], ref)
(score, scores) = cider_scorer.compute_score()
return score, scores
def __init__(self, coco, useBleu=False, useCider=False):
self.coco = coco
self.useBleu = useBleu
self.useCider = useCider
self.params = {'image_id': coco.getImgIds()}
imgIds = self.params['image_id']
gts = {}
for imgId in imgIds:
gts[imgId] = self.coco.imgToAnns[imgId]
if self.useBleu:
self.b_scorer = BleuScorer()
if self.useCider:
self.c_scorer = CiderScorer()
print('tokenization...')
tokenizer = PTBTokenizer()
gts = tokenizer.tokenize(gts)
for imgId in imgIds:
ref = gts[imgId]
assert (type(ref) is list)
assert (len(ref) > 0)
if self.useCider:
self.c_scorer += (None, ref)
if self.useCider:
self.c_scorer.compute_doc_freq()
assert (len(self.c_scorer.ctest) >= max(
self.c_scorer.document_frequency.values()))
def compute_score(self, gts, res):
"""
Main function to compute CIDEr score
: param gts (dict) : {image:tokenized reference sentence}
: param res (dict) : {image:tokenized candidate sentence}
: return: cider (float) : computed CIDEr score for the corpus
"""
cider_scorer = CiderScorer(n=self._n)
for res_id in res:
hypo = res_id['caption']
ref = gts[res_id['image_id']]
# Sanity check.
assert(type(hypo) is list)
assert(len(hypo) == 1)
assert(type(ref) is list)
assert(len(ref) > 0)
cider_scorer += (hypo[0], ref)
(score, scores) = cider_scorer.compute_score(self._df)
return score, scores
def __init__(self, n=4, df="corpus"):
"""
Initialize the CIDEr scoring function
: param n (int): n-gram size
: param df (string): specifies where to get the IDF values from
takes values 'corpus', 'coco-train'
: return: None
"""
# set cider to sum over 1 to 4-grams
self._n = n
self._df = df
self.cider_scorer = CiderScorer(n=self._n, df_mode=self._df)
def setup(self, bottom, top):
if len(bottom) != 2:
raise Exception("Inputs 2 bottom blobs - image_ids and captions.")
if len(top) != 4:
raise Exception("Outputs 3 top blobs - score_weights, input_sentence, target_sentence, mean_score.")
params = ast.literal_eval(self.param_str)
self._end_of_sequence = params['end_of_sequence']
self._ignore_label = params['ignore_label']
# Load vocab
self._vocab = []
with open(params['vocab_path']) as vocab_file:
for word in vocab_file:
self._vocab.append(word.lower().strip())
self._cider = CiderScorer(params['gt_caption_paths'])
def compute_score(self, gts, res):
"""
Main function to compute CIDEr score
:param hypo_for_image (dict) : dictionary with key <image> and value <tokenized hypothesis / candidate sentence>
ref_for_image (dict) : dictionary with key <image> and value <tokenized reference sentence>
:return: cider (float) : computed CIDEr score for the corpus
"""
cider_scorer = CiderScorer(n=self._n, sigma=self._sigma)
for hypo,ref in zip(gts, res):
cider_scorer += (hypo, ref)
(score, scores) = cider_scorer.compute_score()
return score