From: A novel Multi-Layer Attention Framework for visual description prediction using bidirectional LSTM
MODEL | MSVD | MSR VTT | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
B@1 | B@2 | B@3 | B@4 | METEOR | B@1 | B@2 | B@3 | B@4 | METEOR | |
Multi-layer attention (Proposed) without dropout | 70.50 | 56.62 | 49.60 | 33.07 | 51.77 | 60.33 | 43.72 | 34.12 | 19.61 | 39.47 |
Multi-layer attention (Proposed) with dropout | 67.79 | 52.29 | 45.36 | 30.36 | 50.59 | 58.02 | 41.30 | 31.82 | 16.99 | 38.35 |