

Using Deep Learning to Predict Wildfires with Feng Yan - #329
Dec 20, 2019
Feng Yan, an Assistant Professor at the University of Nevada, Reno, is at the forefront of using machine learning for wildfire prediction. He introduces ALERTWildfire, a network of cameras that capture real-time data to enhance monitoring efforts. The conversation dives into innovative camera deployments, the integration of satellite and ground-level data, and overcoming challenges in model training. Feng also discusses leveraging IaaS and FaaS for scalability and cost-effectiveness in tackling the growing threat of wildfires.
AI Snips
Chapters
Transcript
Episode notes
Limitations of Satellite Imagery
- Satellite imagery, while used for wildfire monitoring, has limitations.
- Its low resolution makes it unsuitable for tracking rapidly spreading smoke.
Manual Wildfire Monitoring
- Feng Yan's collaborators deployed cameras to monitor wildfires.
- They manually checked for fires, lacking an automated alert system.
Challenges in Smoke Detection
- Wildfire smoke detection faces challenges due to varied shapes and environmental factors.
- Differentiating smoke from clouds in mountainous areas is difficult, even for humans.