Back to feed
arXiv cs.CL
arXiv cs.CL
7/8/2026
BaFCo: A Document Understanding Benchmark for Complex Bangla Form Comprehension

BaFCo: A Document Understanding Benchmark for Complex Bangla Form Comprehension

Short summary

BaFCo is a new benchmark dataset for evaluating multimodal LLMs on complex Bangla government form comprehension, covering 200 multi-page forms across sectors like agriculture, education, banking, and land management. The benchmark tests Document Layout Analysis and Key Information Extraction using a fine-grained 26-entity annotation schema, evaluating leading MLLMs (ChatGPT, Gemini, Claude, Qwen, Kimi) under zero-shot and chain-of-thought setups. Results reveal significant limitations in current models' ability to accurately localize granular form entities in low-resource Bangla documents.

  • •BaFCo benchmark: 200 complex Bangla government forms for DLA and KIE evaluation
  • •Tests ChatGPT, Gemini, Claude, Qwen, and Kimi MLLMs under zero-shot and chain-of-thought prompts
  • •Current MLLMs show significant limitations in localizing granular entities in Bangla forms

Generated with AI, which can make mistakes.

#research-breakthrough#open-source#ai-tools
Read full article at arXiv cs.CL

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

GSM-Plus-BN: A Perturbation-Based Benchmark for Bangla Mathematical Reasoning in Large Language Models
arXiv cs.CL
DrawingVQA: A Real-World Benchmark for Multi-Depth Visual-Textual Reasoning on Construction Drawings
arXiv CS.AI
Beyond Clean Text: Evaluating Encoder and Decoder Robustness for Bangla Event Detection in Noisy Text
arXiv cs.CL
Indi-RomCoM: Code-Mixed Benchmark for Evaluating LLMs on Romanized Indic-English Instructions
arXiv cs.CL