Zero-shot Image Captioning (ZIC) increasingly utilizes synthetic datasets generated by text-to-image (T2I) models to mitigate the need for costly manual annotation. However, these T2I models often produce images that exhibit semantic misalignments with their corresponding input captions (e.g., missing objects, incorrect attributes), resulting in noisy synthetic image-caption pairs that can hinder model training. Existing dataset pruning techniques are largely designed for removing noisy text in web-crawled data. However, these methods are ill-suited for the distinct challenges of synthetic data, where captions are typically well-formed, but images may be inaccurate representations. To address this gap, we introduce SynC, a novel framework specifically designed to refine synthetic image-caption datasets for ZIC. Instead of conventional filtering or regeneration, SynC focuses on reassigning captions to the most semantically aligned images already present within the synthetic image pool. Our approach employs a one-to-many mapping strategy by initially retrieving multiple relevant candidate images for each caption. We then apply a cycle-consistency-inspired alignment scorer that selects the best image by verifying its ability to retrieve the original caption via image-to-text retrieval. Extensive evaluations demonstrate that SynC consistently and significantly improves performance across various ZIC models on standard benchmarks (MS-COCO, Flickr30k, NoCaps), achieving state-of-the-art results in several scenarios. SynC offers an effective strategy for curating refined synthetic data to enhance ZIC.

MILS (Multimodal Iterative LLM Solver) is a recently published framework that claims "LLMs can see and hear without any training" by leveraging an iterative, LLM-CLIP based approach for zero-shot image captioning. While this MILS approach demonstrates good performance, our investigation reveals that this success comes at a hidden, substantial computational cost due to its expensive multi-step refinement process. In contrast, alternative models such as BLIP-2 and GPT-4V achieve competitive results through a streamlined, single-pass approach. We hypothesize that the significant overhead inherent in MILS's iterative process may undermine its practical benefits, thereby challenging the narrative that zero-shot performance can be attained without incurring heavy resource demands. This work is the first to expose and quantify the trade-offs between output quality and computational cost in MILS, providing critical insights for the design of more efficient multimodal models.

Recently, zero-shot image captioning has gained increasing attention, where only text data is available for training. The remarkable progress in text-to-image diffusion model presents the potential to resolve this task by employing synthetic image-caption pairs generated by this pre-trained prior. Nonetheless, the defective details in the salient regions of the synthetic images introduce semantic misalignment between the synthetic image and text, leading to compromised results. To address this challenge, we propose a novel Patch-wise Cross-modal feature Mix-up (PCM) mechanism to adaptively mitigate the unfaithful contents in a fine-grained manner during training, which can be integrated into most of encoder-decoder frameworks, introducing our PCM-Net. Specifically, for each input image, salient visual concepts in the image are first detected considering the image-text similarity in CLIP space. Next, the patch-wise visual features of the input image are selectively fused with the textual features of the salient visual concepts, leading to a mixed-up feature map with less defective content. Finally, a visual-semantic encoder is exploited to refine the derived feature map, which is further incorporated into the sentence decoder for caption generation. Additionally, to facilitate the model training with synthetic data, a novel CLIP-weighted cross-entropy loss is devised to prioritize the high-quality image-text pairs over the low-quality counterparts. Extensive experiments on MSCOCO and Flickr30k datasets demonstrate the superiority of our PCM-Net compared with state-of-the-art VLMs-based approaches. It is noteworthy that our PCM-Net ranks first in both in-domain and cross-domain zero-shot image captioning. The synthetic dataset SynthImgCap and code are available at https://jianjieluo.github.io/SynthImgCap.

The goal of this paper is to improve the performance of pretrained Vision Transformer (ViT) models, particularly DINOv2, in image clustering task without requiring re-training or fine-tuning. As model size increases, high-norm artifacts anomaly appears in the patches of multi-head attention. We observe that this anomaly leads to reduced accuracy in zero-shot image clustering. These artifacts are characterized by disproportionately large values in the attention map compared to other patch tokens. To address these artifacts, we propose an approach called Inference-Time Attention Engineering (ITAE), which manipulates attention function during inference. Specifically, we identify the artifacts by investigating one of the Query-Key-Value (QKV) patches in the multi-head attention and attenuate their corresponding attention values inside the pretrained models. ITAE shows improved clustering accuracy on multiple datasets by exhibiting more expressive features in latent space. Our findings highlight the potential of ITAE as a practical solution for reducing artifacts in pretrained ViT models and improving model performance in clustering tasks without the need for re-training or fine-tuning.

This paper introduces VLAP, a novel approach that bridges pretrained vision models and large language models (LLMs) to make frozen LLMs understand the visual world. VLAP transforms the embedding space of pretrained vision models into the LLMs' word embedding space using a single linear layer for efficient and general-purpose visual and language understanding. Specifically, we harness well-established word embeddings to bridge two modality embedding spaces. The visual and text representations are simultaneously assigned to a set of word embeddings within pretrained LLMs by formulating the assigning procedure as an optimal transport problem. We predict the assignment of one modality from the representation of another modality data, enforcing consistent assignments for paired multimodal data. This allows vision and language representations to contain the same information, grounding the frozen LLMs' word embedding space in visual data. Moreover, a robust semantic taxonomy of LLMs can be preserved with visual data since the LLMs interpret and reason linguistic information from correlations between word embeddings. Experimental results show that VLAP achieves substantial improvements over the previous linear transformation-based approaches across a range of vision-language tasks, including image captioning, visual question answering, and cross-modal retrieval. We also demonstrate the learned visual representations hold a semantic taxonomy of LLMs, making visual semantic arithmetic possible. Vision-language models (VLMs) have achieved significant progress in demonstrating remarkable transfer and zero-shot capabilities on vision-language downstream tasks (Tan & Bansal, 2019; Lu et al., 2019; Chen et al., 2020; Huang et al., 2020; Radford et al., 2021; Jia et al., 2021).

When trained on large-scale datasets, image captioning models can understand the content of images from a general domain but often fail to generate accurate, detailed captions. To improve performance, pretraining-and-finetuning has been a key strategy for image captioning. However, we find that large-scale bidirectional training between image and text enables zero-shot image captioning. In this paper, we introduce Bidirectional Image Text Training in largER Scale, BITTERS, an efficient training and inference framework for zero-shot image captioning. We also propose a new evaluation benchmark which comprises of high quality datasets and an extensive set of metrics to properly evaluate zero-shot captioning accuracy and societal bias. We additionally provide an efficient finetuning approach for keyword extraction. We show that careful selection of large-scale training set and model architecture is the key to achieving zero-shot image captioning.