Evcap: Retrieval-augmented Image Captioning With External Visual-name Memory For Open-world Comprehension

Li Jiaxuan, Vo Duc Minh, Sugimoto Akihiro, Nakayama Hideki. Arxiv 2023

[Paper]    
Fine Tuning Pretraining Methods Prompting RAG Reinforcement Learning Training Techniques

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.

Similar Work