evaluation measures 0.00201332
automatic evaluation 0.0019211240000000002
human judgements 0.001836756
human judge 0.00178492
human judgement 0.001754397
human judges 0.001744219
single word 0.0016395580000000002
automatic image 0.001638264
evaluation mea 0.001629529
image description 0.001628349
matic evaluation 0.001587046
evaluation script 0.001583746
image descriptions 0.0015727940000000002
test data 0.001547212
description data 0.001507155
only image 0.0014988150000000001
bleu score 0.001480782
original image 0.001439963
data set 0.0013964350000000001
reference sentence 0.0013525759999999999
evaluation 0.00133613
bleu measure 0.00133228
average score 0.0013284780000000001
measures bleu 0.0012854160000000002
automatic measures 0.001262184
precision scores 0.0012544449999999999
candidate sentence 0.001224403
data sets 0.001207603
correlation analysis 0.001202072
test sentences 0.001201052
sentence level 0.001200602
outlier data 0.001180696
data points 0.001180696
rouge score 0.001178264
first sentence 0.00114926
test images 0.001127721
reference descriptions 0.001122748
candidate text 0.001105836
weak correlation 0.001093838
such systems 0.001091342
description task 0.001078524
unigram bleu 0.001069167
natural language 0.001061282
image 0.00105327
sentence pair 0.001049882
description systems 0.001035979
visual aspects 0.001005329
language processing 9.90255E-4
agreement scores 9.86113E-4
visual detectors 9.61655E-4
possible test 9.392229999999999E-4
only matches 9.34106E-4
bleu implementation 8.95044E-4
generation task 8.89679E-4
different auto 8.82863E-4
maximum number 8.79891E-4
description pair 8.756090000000001E-4
recognition task 8.75422E-4
score 8.72556E-4
difficult problem 8.72145E-4
semantic correctness 8.70372E-4
smoothed bleu 8.619000000000001E-4
correlation 8.44988E-4
syntactic tree 8.30322E-4
scription work 8.298679999999999E-4
description pairing 8.26785E-4
description pairings 8.24326E-4
total number 7.84152E-4
main finding 7.83573E-4
words 7.7271E-4
geometric mean 7.70926E-4
arbitrary distance 7.636349999999999E-4
recent approaches 7.617920000000001E-4
sentence 7.49352E-4
brevity penalty 7.3303E-4
prepositional phrases 7.311209999999999E-4
common subse 7.295509999999999E-4
tree substitution 7.26708E-4
measure 7.24054E-4
short translations 7.14667E-4
only yang 6.985419999999999E-4
natural lan 6.898130000000001E-4
guage generation 6.861160000000001E-4
measures 6.7719E-4
language 6.74366E-4
previous studies 6.71787E-4
mechanical turk 6.68015E-4
recent advances 6.669989999999999E-4
scores 6.59188E-4
action recognition 6.55792E-4
penalty factor 6.534150000000001E-4
man judgements 6.30633E-4
mantic correctness 6.23925E-4
certain identification 6.205E-4
problem 6.1684E-4
red scarf 6.155519999999999E-4
partial credit 6.142599999999999E-4
computer vision 6.13932E-4
grammatical correctness 6.102270000000001E-4
chanical turk 6.053230000000001E-4
