EVCap: Retrieval-Augmented Image Captioning with External Visual-Name Memory for Open-World Comprehension

TL;DR

EVCap is a retrieval-augmented image captioning approach that uses external visual-name memory to enable open-world object recognition and description without extensive retraining.

Contribution

It introduces a lightweight, retrievable external memory for object names, allowing LLMs to adapt to new objects efficiently without additional fine-tuning.

Findings

Outperforms other frozen LLM-based methods on benchmarks

Requires only 3.97 million trainable parameters

Effective in out-of-domain and commonsense-violating scenarios

Abstract

Large language models (LLMs)-based image captioning has the capability of describing objects not explicitly observed in training data; yet novel objects occur frequently, necessitating the requirement of sustaining up-to-date object knowledge for open-world comprehension. Instead of relying on large amounts of data and/or scaling up network parameters, we introduce a highly effective retrieval-augmented image captioning method that prompts LLMs with object names retrieved from External Visual--name memory (EVCap). We build ever-changing object knowledge memory using objects' visuals and names, enabling us to (i) update the memory at a minimal cost and (ii) effortlessly augment LLMs with retrieved object names by utilizing a lightweight and fast-to-train model. Our model, which was trained only on the COCO dataset, can adapt to out-of-domain without requiring additional fine-tuning or re-training. Our experiments conducted on benchmarks and synthetic commonsense-violating data show that EVCap, with only 3.97M trainable parameters, exhibits superior performance compared to other methods based on frozen pre-trained LLMs. Its performance is also competitive to specialist SOTAs that require extensive training.