TensorFlow-Based Device Can Detect Wildfire Risk Areas

TensorFlow Fire 1

California just came off of its most destructive wildfire season in history, inspiring two teens to create a device that uses an artificial intelligence program based on and trained with Google’s open-source TensorFlow framework to capture and analyze images of woodland to figure out which areas are most at risk for fires to start and spread. This kind of analysis can help fire departments and nature conservancies to focus their fire prevention efforts, such as pruning and removing dead wood and leaves, on problem areas. It can also help in setting up fire blockades, and getting people on the ground in those areas when reports of fires start coming in, helping to prevent those areas from being overtaken and contributing to wildfires.

The research and development was all done by high school students Aditya Shah and Sanjana Shah. They teamed up with Cal Fire to capture a vast selection of images showing both tame and at-risk woodlands across three counties in California. The two teens created an AI to analyze this data and determine risk factors of a given space based on a picture of it. That AI was built atop Google’s TensorFlow framework, and now inhabits a special device that the duo created. Outfitted with a camera, a solar power panel, and basic computing and networking equipment, the invention is capable of taking and analyzing photos of its surroundings, then sending out its findings.The devices are networked together  to help provide a more cohesive picture of fire risk for a given area.

TensorFlow has found itself an extremely wide range of use cases since going open-source. The framework provides a powerful suite of machine learning tools and examples, making it easy for anybody with basic coding knowledge and access to a powerful enough computer to build and run artificial intelligence programs. TensorFlow is extremely adaptable and extensible, making it an ideal choice for a one-size-fits-all solution that can be reworked for a large number of different applications. It can also scale to just about any size of operation, from a single computer crunching numbers to a complex web of AI nodes working in a cloud-enabled gestalt toward a user’s goals.