The US is home to more than 40 species of bats, but habitat loss, climate change, and disease have taken a toll on populations with many species facing potential extinction. Bats often nest under bridges or overpasses as a way to seek shelter, but lack of awareness about their presence can cause repair projects to unintentionally disrupt or kill groups of these threatened species. To address this issue, a team of researchers from the University of Virginia has created an artificial intelligence (AI) system that can quickly and efficiently detect bat presence without the need for human inspections.
Using a pool of digital photographs of bridges with and without signs of bats, the researchers taught an AI model to recognize the features and traits that identify the presence of bats. Bats are primarily identified by the presence of guano, or excrement, but it can be difficult for the untrained eye to discern between guano and other structural stains like water seeps, rust, or asphalt leaching. Using the new AI system, officials and workers can upload photos of a site and quickly know if there are signs of bat presence.
Bats roost in groups, often in the thousands, so ensuring they are not using a bridge as a home before starting construction is critical for conserving their populations. Although sometimes spooky, bats play an important role in ecosystems by pollinating plants, spreading seeds, and keeping insect populations in check.
Moving forwards, the Virginia Department of Transportation plans to conduct a pilot study for bridge inspectors and environmental staff to test out the AI system potential construction sites. If effective in a real world setting, this system could be used to protect bats all around the world.
Source study: Transportation Research Record – Deep Learning-Based Visual Identification of Signs of Bat Presence in Bridge Infrastructure Images: A Transfer Learning Approach