Hateful memes have emerged as a particularly challenging form of online abuse, motivating the development of automated detection systems. Most prior approaches rely on direct detection, producing only binary predictions. Such models fail to provide the context and explanations that real-world moderation requires. Recent Explain-then-Detect approaches, using Chain-of-Thought prompting or LMM agents, perform worse than simple SFT baselines, and even advanced post-training methods such as GRPO fail to close the gap. Our analysis identifies two key issues of such systems: important policy-relevant cues such as targets and attack types are not hypothesized by the model as a likely explanation; and the binary reward signal is insufficient to guide reasoning. To address these challenges, we propose ExPO-HM (Explain-then-Detect Policy Optimization for Hateful Memes), inspired by the training and evaluation process of human annotators. ExPO-HM combines SFT warmup, GRPO with curriculum learning, and Conditional Decision Entropy (CDE) as both metric and reward for reasoning quality. Across three hateful meme benchmarks, ExPO-HM achieves state-of-the-art performance on binary detection, fine-grained classification, and reasoning quality, with up to 15\% and 17\% F1 improvement over the GRPO and DPO baselines, respectively. By moving hateful meme detection from simple binary alarms to explanation-driven detection, ExPO-HM provides accurate, interpretable, and actionable moderation support.

Hateful memes have become a significant concern on the Internet, necessitating robust automated detection systems. While large multimodal models have shown strong generalization across various tasks, they exhibit poor generalization to hateful meme detection due to the dynamic nature of memes tied to emerging social trends and breaking news. Recent work further highlights the limitations of conventional supervised fine-tuning for large multimodal models in this context. To address these challenges, we propose Large Multimodal Model Retrieval-Guided Contrastive Learning (LMM-RGCL), a novel two-stage fine-tuning framework designed to improve both in-domain accuracy and cross-domain generalization. Experimental results on six widely used meme classification datasets demonstrate that LMM-RGCL achieves state-of-the-art performance, outperforming agent-based systems such as VPD-PALI-X-55B. Furthermore, our method effectively generalizes to out-of-domain memes under low-resource settings, surpassing models like GPT-4o.

Hateful meme detection presents a significant challenge as a multimodal task due to the complexity of interpreting implicit hate messages and contextual cues within memes. Previous approaches have fine-tuned pre-trained vision-language models (PT-VLMs), leveraging the knowledge they gained during pre-training and their attention mechanisms to understand meme content. However, the reliance of these models on implicit knowledge and complex attention mechanisms renders their decisions difficult to explain, which is crucial for building trust in meme classification. In this paper, we introduce IntMeme, a novel framework that leverages Large Multimodal Models (LMMs) for hateful meme classification with explainable decisions. IntMeme addresses the dual challenges of improving both accuracy and explainability in meme moderation. The framework uses LMMs to generate human-like, interpretive analyses of memes, providing deeper insights into multimodal content and context. Additionally, it uses independent encoding modules for both memes and their interpretations, which are then combined to enhance classification performance. Our approach addresses the opacity and misclassification issues associated with PT-VLMs, optimizing the use of LMMs for hateful meme detection. We demonstrate the effectiveness of IntMeme through comprehensive experiments across three datasets, showcasing its superiority over state-of-the-art models.

The prevalence of multi-modal content on social media complicates automated moderation strategies. This calls for an enhancement in multi-modal classification and a deeper understanding of understated meanings in images and memes. Although previous efforts have aimed at improving model performance through fine-tuning, few have explored an end-to-end optimization pipeline that accounts for modalities, prompting, labeling, and fine-tuning. In this study, we propose an end-to-end conceptual framework for model optimization in complex tasks. Experiments support the efficacy of this traditional yet novel framework, achieving the highest accuracy and AUROC. Ablation experiments demonstrate that isolated optimizations are not ineffective on their own.

Recent advances show that two-stream approaches have achieved outstanding performance in hateful meme detection. However, hateful memes constantly evolve as new memes emerge by fusing progressive cultural ideas, making existing methods obsolete or ineffective. In this work, we explore the potential of Large Multimodal Models (LMMs) for hateful meme detection. To this end, we propose Evolver, which incorporates LMMs via Chain-of-Evolution (CoE) Prompting, by integrating the evolution attribute and in-context information of memes. Specifically, Evolver simulates the evolving and expressing process of memes and reasons through LMMs in a step-by-step manner. First, an evolutionary pair mining module retrieves the top-k most similar memes in the external curated meme set with the input meme. Second, an evolutionary information extractor is designed to summarize the semantic regularities between the paired memes for prompting. Finally, a contextual relevance amplifier enhances the in-context hatefulness information to boost the search for evolutionary processes. Extensive experiments on public FHM, MAMI, and HarM datasets show that CoE prompting can be incorporated into existing LMMs to improve their performance. More encouragingly, it can serve as an interpretive tool to promote the understanding of the evolution of social memes.

The digital revolution and the advent of the world wide web have transformed human communication, notably through the emergence of memes. While memes are a popular and straightforward form of expression, they can also be used to spread misinformation and hate due to their anonymity and ease of use. In response to these challenges, this paper introduces a solution developed by team 'Baseline' for the AI Singapore Online Safety Prize Challenge. Focusing on computational efficiency and feature engineering, the solution achieved an AUROC of 0.76 and an accuracy of 0.69 on the test dataset. As key features, the solution leverages the inherent probabilistic capabilities of large Vision-Language Models (VLMs) to generate task-adapted feature encodings from text, and applies a distilled quantization tailored to the specific cultural nuances present in Singapore. This type of processing and fine-tuning can be adapted to various visual and textual understanding and classification tasks, and even applied on private VLMs such as OpenAI's GPT. Finally it can eliminate the need for extensive model training on large GPUs for resource constrained applications, also offering a solution when little or no data is available.

Internet memes have become a powerful means for individuals to express emotions, thoughts, and perspectives on social media. While often considered as a source of humor and entertainment, memes can also disseminate hateful content targeting individuals or communities. Most existing research focuses on the negative aspects of memes in high-resource languages, overlooking the distinctive challenges associated with low-resource languages like Bengali (also known as Bangla). Furthermore, while previous work on Bengali memes has focused on detecting hateful memes, there has been no work on detecting their targeted entities. To bridge this gap and facilitate research in this arena, we introduce a novel multimodal dataset for Bengali, BHM (Bengali Hateful Memes). The dataset consists of 7,148 memes with Bengali as well as code-mixed captions, tailored for two tasks: (i) detecting hateful memes, and (ii) detecting the social entities they target (i.e., Individual, Organization, Community, and Society). To solve these tasks, we propose DORA (Dual cO attention fRAmework), a multimodal deep neural network that systematically extracts the significant modality features from the memes and jointly evaluates them with the modality-specific features to understand the context better. Our experiments show that DORA is generalizable on other low-resource hateful meme datasets and outperforms several state-of-the-art rivaling baselines.

Recently, large language models (LLMs) have taken the spotlight in natural language processing. Further, integrating LLMs with vision enables the users to explore more emergent abilities in multimodality. Visual language models (VLMs), such as LLaVA, Flamingo, or GPT-4, have demonstrated impressive performance on various visio-linguistic tasks. Consequently, there are enormous applications of large models that could be potentially used on social media platforms. Despite that, there is a lack of related work on detecting or correcting hateful memes with VLMs. In this work, we study the ability of VLMs on hateful meme detection and hateful meme correction tasks with zero-shot prompting. From our empirical experiments, we show the effectiveness of the pretrained LLaVA model and discuss its strengths and weaknesses in these tasks. Warning: This paper contains examples of hate speech.

The recently introduced hateful meme challenge demonstrates the difficulty of determining whether a meme is hateful or not. Specifically, both unimodal language models and multimodal vision-language models cannot reach the human level of performance. Motivated by the need to model the contrast between the image content and the overlayed text, we suggest applying an off-the-shelf image captioning tool in order to capture the first. We demonstrate that the incorporation of such automatic captions during fine-tuning improves the results for various unimodal and multimodal models. Moreover, in the unimodal case, continuing the pre-training of language models on augmented and original caption pairs, is highly beneficial to the classification accuracy.