Let there be a clock on the beach: Reducing Object Hallucination in Image Captioning

Explaining an image with missing or non-existent objects is known as object bias (hallucination) in image captioning. This behaviour is quite common in the state-of-the-art captioning models which is not desirable by humans. To decrease the object hallucination in captioning, we propose three simple yet efficient training augmentation method for sentences which requires no new training data or increase in the model size. By extensive analysis, we show that the proposed methods can significantly diminish our models' object bias on hallucination metrics. Moreover, we experimentally demonstrate that our methods decrease the dependency on the visual features. All of our code, configuration files and model weights are available online(1).

Automated summary

The article discusses object bias (hallucination) in image captioning and presents three simple yet efficient training augmentation methods to reduce it without the need for new data or increased model size. The proposed methods are shown to significantly decrease object bias in the models based on hallucination metrics, and reduce dependency on visual features through experimental demonstration. All code, configuration files, and model weights are available online.