Share your thoughts, 1 month free Claude Pro on usSee more

Image Question Answering on DAQUAR REDUCED (test)

60.3Accuracy

Human

Updated 1mo ago

Evaluation Results

Method	Links
Human 2015.11		60.3	61	79
Human Baseline 2016.03		60.27	61.04	78.96
SAN(2, LSTM) 2015.11		46.2	51.2	85.1
A+C+Selected-K-LSTM 2016.03		46.13	51.83	83.95
Att+Cap+Know-LSTM 2016.03		45.79	51.53	83.91
SAN(2, CNN) 2015.11		45.5	50.2	83.6
Yang et al. 2016.03		45.5	50.2	83.6
SAN(1, LSTM) 2015.11		45.2	49.6	84
SAN(1, CNN) 2015.11		45.2	49.6	83.7
Att+Cap-LSTM 2016.03		44.78	50.07	83.85
Noh et al. 2016.03		44.48	49.56	83.95
Att+Know-LSTM 2016.03		41.08	46.04	82.39
Cap+Know-LSTM 2016.03		40.81	45.04	82.01
Xu et al. 2016.03		40.07	-	-
Att-LSTM 2016.03		40.07	45.43	82.67
IMG-CNN 2015.11		39.7	44.9	83.1
Ma et al. 2016.03		39.66	44.86	83.06
VggNet+ft-LSTM 2016.03		39.13	44.03	83.33
VggNet-LSTM 2016.03		38.72	43.97	83.01
2-VIS+BLSTM 2015.11		35.8	46.8	82.2
2-VIS+BLSTM 2016.03		35.78	46.83	82.15
Language + IMG 2015.11		34.7	40.8	79.5
Askneuron 2016.03		34.68	40.76	79.54
VIS+LSTM 2016.03		34.41	46.05	82.23
VIS+LSTM 2015.11		34.4	46.1	82.2
IMG+BOW 2015.11		34.2	45	81.5
VIS+BOW 2016.03		34.17	44.99	81.48
BOW 2015.11		32.7	43.2	81.3
LSTM 2015.11		32.7	43.5	81.6
Language 2015.11		31.7	38.4	80.1
GUESS 2016.03		18.24	29.65	77.59
GUESS 2015.11		18.2	29.7	77.6
Multi-World 2015.11		12.7	18.2	51.5