English

An Efficient Modern Baseline for FloodNet VQA

Computer Vision and Pattern Recognition 2022-05-31 v1 Artificial Intelligence Computation and Language

Abstract

Designing efficient and reliable VQA systems remains a challenging problem, more so in the case of disaster management and response systems. In this work, we revisit fundamental combination methods like concatenation, addition and element-wise multiplication with modern image and text feature abstraction models. We design a simple and efficient system which outperforms pre-existing methods on the FloodNet dataset and achieves state-of-the-art performance. This simplified system requires significantly less training and inference time than modern VQA architectures. We also study the performance of various backbones and report their consolidated results. Code is available at https://github.com/sahilkhose/floodnet_vqa.

Keywords

Cite

@article{arxiv.2205.15025,
  title  = {An Efficient Modern Baseline for FloodNet VQA},
  author = {Aditya Kane and Sahil Khose},
  journal= {arXiv preprint arXiv:2205.15025},
  year   = {2022}
}

Comments

Under review, 4 pages, 2 figures, 1 table

R2 v1 2026-06-24T11:32:59.587Z