HyperAI

Image Captioning On Nocaps Near Domain

Métriques

B1
B2
B3
B4
CIDEr
METEOR
ROUGE-L
SPICE

Résultats

Résultats de performance de divers modèles sur ce benchmark

Tableau comparatif
Nom du modèleB1B2B3B4CIDErMETEORROUGE-LSPICE
Modèle 174.7753.6730.6613.8561.9822.5549.459.83
clipcap-clip-prefix-for-image-captioning----66.82--10.92
Modèle 379.5960.5238.9520.7279.4425.6453.1811.88
git-a-generative-image-to-text-transformer88.975.8658.938.95125.5132.9563.6616.11
Modèle 584.4769.6651.9533.46109.3331.0860.3414.79
Modèle 673.654.2634.5918.9563.9624.5251.2311.14
Modèle 773.1953.5632.9417.4950.3422.4349.799.7
Modèle 880.5462.3240.6522.3782.0725.9154.7811.53
Modèle 973.6954.132.3715.9953.2121.9349.639.26
Modèle 1080.2462.3141.0721.5380.2125.9854.5212.12
Modèle 1181.9365.8846.7227.9489.8727.8957.3412.98
Modèle 1272.9153.7433.4918.0458.523.1250.5310.28
pali-a-jointly-scaled-multilingual-language-------15.75
Modèle 1487.5374.4957.8938.92120.7332.7162.9115.54
Modèle 1577.7659.038.2921.076.3425.5953.1511.87
Modèle 1679.2162.2640.7722.5684.026.354.6212.47
Modèle 1779.2162.0642.5125.0679.1426.8755.2412.14
Modèle 1875.2556.9336.9120.4956.8523.651.8410.33
Modèle 1975.2556.9336.9120.4956.8523.651.8410.33
clipcap-clip-prefix-for-image-captioning----67.69--11.26
Modèle 2177.0556.9736.8419.8584.5828.4253.0614.72
Modèle 2270.8450.7930.2616.1447.5321.4848.619.28
Modèle 2370.0548.9226.1912.1140.4520.0547.048.28
pali-a-jointly-scaled-multilingual-language88.5775.5658.9939.98124.3533.4763.9915.75
git-a-generative-image-to-text-transformer88.5675.4858.4638.44123.9232.8663.515.96
Modèle 2674.0755.5335.2218.7949.6222.4150.779.54
Modèle 2781.6266.6549.3931.4299.5130.4858.8314.88
Modèle 2879.6960.7539.0620.9779.7225.6453.3711.81
Modèle 2984.0468.5849.9830.78100.1529.5359.2314.15
Modèle 3082.5566.5547.829.095.6929.1158.2214.37
Modèle 3168.8648.726.8512.646.6420.1847.138.37
Modèle 3273.7353.9833.117.2851.1622.2750.09.7
vinvl-making-visual-representations-matter-in82.7766.9447.0227.9795.1628.2457.9513.36
Modèle 3477.6858.3137.0419.8574.224.9752.6411.45
simvlm-simple-visual-language-model84.3669.8352.4233.74110.7630.9760.4614.61
Modèle 3683.5867.9949.2929.9695.7328.8458.4713.64
vivo-surpassing-human-performance-in-novel86.4872.655.2636.31115.5431.861.915.06
Modèle 3879.6762.7342.8724.885.8926.6855.3712.24
Modèle 3983.7168.5650.932.72108.0430.7959.814.71
Modèle 4079.6163.0143.5925.8587.4126.6355.6312.11