Semantica@DravidianLangTech 2026: Vision-Language Models for Hierarchical Political Meme Classification in Tamil and Malayalam

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Social Sciences & Behavioral Studies · Depth: Expert, quick

Summary

Semantica@DravidianLangTech 2026 explored Vision-Language Models (VLMs) for hierarchical political meme classification in low-resource languages, specifically Tamil and Malayalam. This initiative addressed the challenge of detecting political trolling in these languages, which suffer from limited datasets and tools. The research evaluated various approaches, including text-only models like IndicBERTv2 and XLM-RoBERTa, classical multimodal fusion using EfficientNet, and several Qwen-VL models. Among the submitted systems, Qwen2.5-VL-7B-Instruct, fine-tuned with 4-bit QLoRA, achieved notable results, securing 3rd place in the Malayalam track and 4th in the Tamil track based on weighted-F1 score. Post-evaluation experiments with Qwen3-VL-8B further enhanced macro-F1 performance, underscoring the efficacy of VLMs for multilingual political meme classification in low-resource contexts.

Key takeaway

For NLP Engineers developing content moderation systems in low-resource languages, this research indicates that Vision-Language Models are highly effective. You should prioritize exploring Qwen-VL models, specifically Qwen2.5-VL-7B-Instruct or Qwen3-VL-8B, fine-tuned with 4-bit QLoRA, for hierarchical political meme classification. This approach offers a robust solution for detecting political trolling in languages like Tamil and Malayalam, where traditional methods struggle due to data scarcity.

Key insights

Vision-Language Models effectively classify hierarchical political memes in low-resource Dravidian languages.

Principles

• VLMs outperform text-only and classical multimodal fusion.
• QLoRA fine-tuning enhances VLM performance.
• Low-resource languages benefit from advanced VLMs.

Method

The approach involved evaluating text-only, classical multimodal fusion, and VLM architectures, with QLoRA fine-tuning applied to Qwen-VL models for hierarchical classification.

In practice

• Apply Qwen-VL models for multilingual meme analysis.
• Use 4-bit QLoRA for VLM fine-tuning.
• Target low-resource language content moderation.

Topics

• Vision-Language Models
• Political Meme Classification
• Low-Resource NLP
• Tamil Malayalam
• Qwen-VL
• QLoRA Fine-tuning

Best for: Research Scientist, AI Scientist, NLP Engineer

Related on AIssential

• From Codebooks to VLMs: Evaluating Automated Visual Discourse Analysis for Climate Change on Social Media — VLMs can effectively analyze social media images for climate discourse, even with moderate per-image accuracy.
• Fine-Tuning Qwen3-VL — Fine-tuning Qwen3-VL for image-to-HTML requires careful dataset preparation and context length management.
• MemLens: Benchmarking Multimodal Long-Term Memory in Large Vision-Language Models — MEMLENS benchmarks multimodal long-term memory, revealing distinct strengths and weaknesses in LVLMs and memory-augmented agents.
• Comparing Qwen3-VL AI Models for OCR Task — Quantized larger Qwen3-VL models can outperform smaller, unquantized versions in both speed and accuracy for complex OCR.
• Merve Noyan: The Future of Vision in ML - HF Podcast #1 — Vision models are maturing, shifting from architectural breakthroughs to optimization and integration with language for real-world applications.
• Building an LLM fine-tuning Dataset — Building multi-speaker Reddit datasets for LLM fine-tuning is complex but achievable, with efficient QLoRA methods requiring minimal samples.

Open in AIssential →

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