import { PageContainer, Heading, Button } from '@ifrc-go/ui'; import { useNavigate } from 'react-router-dom'; import { useFilterContext } from '../hooks/useFilterContext'; import styles from './HelpPage.module.css'; export default function HelpPage() { const navigate = useNavigate(); const { setShowReferenceExamples } = useFilterContext(); const handleUploadNow = () => { navigate('/upload'); }; const handleSeeExamples = () => { setShowReferenceExamples(true); navigate('/explore'); }; const handleViewVlmDetails = () => { navigate('/analytics?view=crisis_maps'); }; return (
Introduction
In collaboration with the IFRC, PromptAid Vision is a tool that generates textual descriptions of crisis maps/crisis drone images utiliing Visual language models. This prototype is for collecting data for the fine-tuning of our own models. We aim to utilize AI tools to support national societies with rapid decision making during emergencies.
Guidelines
To make the process smoother, please follow the guidelines below:
  • Avoid uploading images that are not crisis maps/crisis drone images.
  • Confirm the image details prior to modifying the description.
  • Before the modification, please read the description generated and provide a rating via the rating sliders.
  • Click the "Submit" button to save the description.
VLMs
PromptAid Vision uses a variety of Visual Language Models (VLMs). A random VLM is selected for each upload. Therefore feel free to delete and reupload. You can view performance details here:
Dataset
All users are able to export the dataset. You could apply filters when exporting, and it have the option to organize based on model fine-tuning formats.
Contact us
Need help or have questions about PromptAid Vision? Our team is here to support you.
); }