SigJBS@LT-EDI 2026: Multimodal Homophobia and Transphobia Meme Classification

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Natural Language Processing, Computer Vision · Depth: Expert, medium

Summary

The SigJBS@LT-EDI 2026 system addresses the challenging task of classifying homophobia and transphobia in multimodal memes across English, Hindi, and Chinese languages. This system employs a lightweight and reproducible pipeline that integrates CLIP ViT-L/14 visual embeddings, EasyOCR for text extraction, and TF–IDF lexical features, all fed into a multinomial logistic regression classifier. For enhanced performance, it optionally incorporates a LoRA-adapted Qwen2-VL model and a CLIP zero-shot classifier, combining their predictions via weighted majority voting. The approach demonstrates that robust multilingual meme moderation can be achieved using strong pretrained transfer features and explicit OCR, minimizing the need for extensive fine-tuning. On the official LT-EDI@ACL 2026 leaderboard, the system secured 1st place in Hindi, 3rd in English, and 5th in Chinese.

Key takeaway

For NLP Engineers developing multilingual meme moderation systems, you should prioritize lightweight architectures leveraging pretrained multimodal models. Your approach can achieve robust performance across languages like Hindi, English, and Chinese by integrating explicit OCR and combining diverse expert module predictions. This strategy minimizes extensive fine-tuning while maintaining high accuracy, as demonstrated by top rankings on the LT-EDI@ACL 2026 leaderboard. Consider implementing a similar pipeline to efficiently scale your moderation efforts.

Key insights

Robust multilingual meme moderation is achievable with pretrained multimodal features and explicit OCR, minimizing fine-tuning.

Principles

• Multimodal interaction is key for meme meaning.
• Pretrained features enable robust multilingual moderation.
• Explicit OCR enhances text extraction in memes.

Method

A multimodal pipeline combines CLIP ViT-L/14 visual embeddings, EasyOCR text, TF–IDF features, and a multinomial logistic regression classifier. Optional expert modules (LoRA-adapted Qwen2-VL, CLIP zero-shot) use weighted majority voting.

In practice

• Integrate CLIP ViT-L/14 for visual understanding.
• Use EasyOCR for multilingual text extraction.
• Combine predictions via weighted majority voting.

Topics

• Multimodal Classification
• Meme Detection
• Homophobia Detection
• Transphobia Detection
• Multilingual NLP
• Content Moderation

Best for: Research Scientist, AI Scientist, NLP Engineer, Computer Vision Engineer

Related on AIssential

• Multi-Stage Training for Abusive Comment Detection in Indic Languages — Multi-stage training with language preprocessing and model ensembles improves abusive comment detection while minimizing false positives.
• TeamHerald@CHIPSAL 2026: Hate Speech Detection and Sentiment Analysis of Nepali Memes using Transformer-based Architectures and Ensemble Learning — Ensemble strategies for meme analysis perform differently for binary vs. multi-class tasks, requiring task-specific aggregation.
• Comparison of Modern Multilingual Text Embedding Techniques for Hate Speech Detection Task — Supervised models with multilingual embeddings and gradient boosted trees excel at hate speech detection.
• A Multimodal Framework for Financial Fake News Detection for Brazilian Portuguese — Multimodal analysis, integrating text and image-extracted text, significantly improves financial fake news detection.
• From Comments to Trends: Studi NLP dan Computational Linguistics dalam Mendeteksi Topik Viral dari… — NLP and computational linguistics are crucial for detecting social media trends from vast, informal, and diverse comment data.
• mdok-style at SemEval-2026 Task 9: Finetuning LLMs for Multilingual Polarization Detection — Finetuning mid-sized LLMs with QLoRA and augmented data improves multilingual online polarization detection.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.