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

Image Captioning Evaluation on Nebula

57.5Kendall tau_c

RIGEL

Updated 25d ago

Evaluation Results

Method
RIGEL 2026.06	57.5	60.6
BEiTScore-L/14 2026.05	56.7	-
BEiTScore-L/14 w/o Stage 1 2026.05	56.3	-
Pearl 2025.12	55.4	58.4
Pearl 2025.12	55.4	58.4
Pearl 2026.06	55.4	58.4
BEiTScore-B/32 2026.05	55.3	-
DENEB 2025.12	55.1	58.1
DENEB 2026.06	55.1	58.1
RIGEL 2026.06	55.1	58.1
Polos 2025.12	55	58
Polos 2026.06	55	58
EXPERT 2025.12	54.9	-
EXPERT 2026.05	54.9	-
BEiTScore-B/32 w/o Stage 1 2026.05	54.6	-
Pomms 2026.04	54.1	57
Pomms 2026.04	54.1	57
Pomms 2026.04	53.7	56.6
Pomms 2026.04	53.6	56.6
Pomms 2026.04	53.5	56.4
Pomms 2026.04	53.5	56.4
GPT-4o 2026.04	53.2	56.1
Pearl 2025.12	53	55.9
Pearl 2025.12	53	55.9
Pearl 2026.06	53	55.9
FLEUR 2026.05	52.7	-
Qwen2.5-VL-7B 2026.04	52.4	55.3
RefPAC-S 2025.12	51.9	54.7
RefPAC-S 2026.06	51.9	54.7
InternVL2.5-8B 2026.04	51.3	54.1
RefCLIP-S 2025.12	50.8	53.6
RefCLIP-S 2026.06	50.8	53.6
BLIP2Score 2025.12	50.7	53
BLIP2-Score 2026.06	50.7	53
RefPAC-S++ 2025.12	50.6	53.3
RefPAC-S++ 2026.06	50.6	53.3
SPARCS 2025.12	50	53.9
VQAScore (CLIP-FlanT5-XXL) 2026.05	49.4	-
LongCLIP-S/ViT-B 2026.05	48.9	-
PAC-S++/ViT-L 2026.05	48.9	-
CIDEr 2025.12	48.8	51.5
CLIP-S/ViT-L 2026.05	48.8	-
CIDEr 2026.06	48.8	51.5
LongCLIP-S/ViT-L 2026.05	48.6	-
PAC-S 2025.12	48.3	51
PAC-S 2026.06	48.3	51
CLIP-S 2025.12	47.9	50.5
PAC-S++ 2025.12	47.9	50.5
PAC-S/ViT-L 2026.05	47.9	-
CLIP-S 2026.06	47.9	50.5
PAC-S++ 2026.06	47.9	50.5
CLIP-S/ViT-B 2026.05	47.8	-
METEOR 2025.12	47.6	50.2
METEOR 2026.06	47.6	50.2
SPICE 2025.12	47.4	51.5
SPICE 2026.06	47.4	51.5
PAC-S/ViT-B 2026.05	47.2	-
BERTScore 2025.12	47.1	47.5
BERTScore 2026.06	47.1	47.5
PAC-S++/ViT-B 2026.05	46.9	-
BARTScore 2025.12	45	47.4
BLEU 2025.12	44.1	46.5
BLEU 2026.06	44.1	46.5
ROUGE 2025.12	43.4	45.8
ROUGE 2026.06	43.4	45.8
SPECS 2026.05	34.4	-