Multi-Object Hallucination in Vision-Language Models
This work addresses a critical issue for improving the reliability of vision-language models in realistic visual scenes, though it is incremental as it builds on existing benchmarks for object hallucination.
The paper tackles the problem of multi-object hallucination in large vision-language models, where models invent nonexistent objects when focusing on multiple objects, and introduces ROPE, an automated evaluation protocol that found LVLMs suffer more hallucinations with multiple objects, with hallucination rates affected by object class distribution and other factors.
Large vision language models (LVLMs) often suffer from object hallucination, producing objects not present in the given images. While current benchmarks for object hallucination primarily concentrate on the presence of a single object class rather than individual entities, this work systematically investigates multi-object hallucination, examining how models misperceive (e.g., invent nonexistent objects or become distracted) when tasked with focusing on multiple objects simultaneously. We introduce Recognition-based Object Probing Evaluation (ROPE), an automated evaluation protocol that considers the distribution of object classes within a single image during testing and uses visual referring prompts to eliminate ambiguity. With comprehensive empirical studies and analysis of potential factors leading to multi-object hallucination, we found that (1). LVLMs suffer more hallucinations when focusing on multiple objects compared to a single object. (2). The tested object class distribution affects hallucination behaviors, indicating that LVLMs may follow shortcuts and spurious correlations. (3). Hallucinatory behaviors are influenced by data-specific factors, salience and frequency, and model intrinsic behaviors. We hope to enable LVLMs to recognize and reason about multiple objects that often occur in realistic visual scenes, provide insights, and quantify our progress towards mitigating the issues.