ViBe: A Text-to-Video Benchmark for Evaluating Hallucination in Large Multimodal Models
This provides a critical resource for evaluating T2V reliability and advancing hallucination detection, addressing a key issue for users of large multimodal models.
The paper tackles the problem of hallucinated content in Text-to-Video (T2V) models by introducing ViBe, a large-scale benchmark dataset of 3,782 hallucinated videos from ten models, identifying five hallucination types and establishing a baseline classification with 0.345 accuracy and 0.342 F1 score.
Recent advances in Large Multimodal Models (LMMs) have expanded their capabilities to video understanding, with Text-to-Video (T2V) models excelling in generating videos from textual prompts. However, they still frequently produce hallucinated content, revealing AI-generated inconsistencies. We introduce ViBe (https://vibe-t2v-bench.github.io/): a large-scale dataset of hallucinated videos from open-source T2V models. We identify five major hallucination types: Vanishing Subject, Omission Error, Numeric Variability, Subject Dysmorphia, and Visual Incongruity. Using ten T2V models, we generated and manually annotated 3,782 videos from 837 diverse MS COCO captions. Our proposed benchmark includes a dataset of hallucinated videos and a classification framework using video embeddings. ViBe serves as a critical resource for evaluating T2V reliability and advancing hallucination detection. We establish classification as a baseline, with the TimeSFormer + CNN ensemble achieving the best performance (0.345 accuracy, 0.342 F1 score). While initial baselines proposed achieve modest accuracy, this highlights the difficulty of automated hallucination detection and the need for improved methods. Our research aims to drive the development of more robust T2V models and evaluate their outputs based on user preferences.