Wildlife Image Retrieval System Flaws Exposed by Ecological Researchers

This article discusses a research project called “INQUIRE” that aims to develop a query system for retrieving images from large collections of biodiversity data. Here are the main points:

Background:

• Biodiversity datasets are growing rapidly, making it difficult for scientists to review and analyze them manually.

• The researchers aim to address this challenge by developing a query system that can efficiently search through large image collections.

INQUIRE Project:

• The INQUIRE project is a text-to-image retrieval benchmark that allows users to filter searches by species, enabling quicker discovery of relevant results.

• The system uses artificial intelligence (AI) and machine learning (ML) algorithms to analyze the images and provide accurate results.

Key Findings:

• Some vision models are already precise enough to aid wildlife scientists with retrieving some images, but many tasks are still too difficult for even the largest, best-performing models.

• The INQUIRE system has shown promising results in uncovering secondary data, which is critical for fundamental science and real-world impacts in ecology and conservation.

Future Work:

• The researchers plan to improve the re-ranking system by augmenting current models to provide better results.

• They will also work with iNaturalist to develop a query system that can help scientists and other curious minds find the images they actually want to see.

Related Research:

• Other related research projects mentioned in the article include:

  • Advancing urban tree monitoring with AI-powered digital twins

  • CSAIL’s Sara Beery, Marzyeh Ghassemi, and Yoon Kim earn AI2050 Early Career Fellowships

People Involved:

• Edward Vendrow (lead author)

• Sara Beery (co-author)

• Omiros Pantazis (co-author)

• Alexander Shepard (iNaturalist software engineer)

• Gabriel Brostow (University College London professor)

• Kate Jones (University College London professor)

• Oisin Mac Aodha (University of Edinburgh associate professor)

• Grant Van Horn (University of Massachusetts at Amherst Assistant Professor)

Support:

• The project was supported by various organizations, including:

  • Generative AI Laboratory at the University of Edinburgh

  • U.S. National Science Foundation/Natural Sciences and Engineering Research Council of Canada Global Center on AI and Biodiversity Change

  • Royal Society Research Grant

  • Biome Health Project funded by the World Wildlife Fund United Kingdom