List your company

Object Hallucination Detection on nocaps FOIL In-Domain

88.8APEXPERT
66.12872.01477.983.786May 20, 2026

Evaluation results

MethodLinks
EXPERTMethod Category=LLM-based2026.0588.891.797.6
VQAScoreMethod Category=LLM-based, Backbone=CLIP-FlanT5-XXL2026.0588.590.295.9
BEiTScore-L/14Method Category=Encoder-based, Backbone=BEiT-L/142026.0587.990.797.1
BEiTScore-B/32Method Category=Encoder-based, Backbone=BEiT-B/322026.0585.887.798
FLEURMethod Category=LLM-based2026.0582.586.696.3
LongCLIP-SMethod Category=Encoder-based, Backbone=ViT-L2026.0574.976.897.1
PAC-SMethod Category=Encoder-based, Backbone=ViT-L2026.0573.57791.4
PAC-S++Method Category=Encoder-based, Backbone=ViT-L2026.0573.377.895.5
SPECSMethod Category=Encoder-based2026.0573.176.598
LongCLIP-SMethod Category=Encoder-based, Backbone=ViT-B2026.0572.876.193.1
PAC-S++Method Category=Encoder-based, Backbone=ViT-B2026.0571.974.790.6
ALOHaMethod Category=LLM-based2026.0571.8——
BEiTScore-L/14 w/o Stage 1Method Category=Encoder-based, Backbone=BEiT-L/14, Training Stage=w/o Stage 12026.0571.476.394.3
PAC-SMethod Category=Encoder-based, Backbone=ViT-B2026.0571.374.689.4
BEiTScore-B/32 w/o Stage 1Method Category=Encoder-based, Backbone=BEiT-B/32, Training Stage=w/o Stage 12026.057072.994.3
CLIP-SMethod Category=Encoder-based, Backbone=ViT-L2026.056971.887.8
CLIP-SMethod Category=Encoder-based, Backbone=ViT-B2026.05677084.1
GPT4o (API)Method Category=LLM-based2026.05——94.7