I would like to share this donation campaign to raise donations to support AI For Mankind’s early wildfire detection efforts. The mission of the wildfire detection system is to detect wildfires as early as possible when they are still small and therefore easier to put out.
On Nov 29 2018, a few of AI For Mankind’s members were sitting at Blue Bottle Coffee shop brainstorming and discussing what we could do to help solve wildfire crisis in California after experiencing one of the worst wildfire seasons in 2018. I remembered that masks and air purifiers were…
Apple Fire broke out on Sunday evening July 31. It has since turned into a massive wildfire burning in Riverside and San Bernardino counties. As of Aug 2 ~9:25 PM, it has scorched 20,516 acres with 5% containment. https://inciweb.nwcg.gov/incident/6902/
AI For Mankind’s Wildfire Smoke Detector managed to detect the smoke…
AI For Mankind has been running and evaluating its latest wildfire smoke detection system scanning mountain top pictures captured by HPWREN cameras around Southern California since last few weeks. Along the way, we gained a glimpse into different interesting false positives.
We noticed that lens flares, lens dirts, and flying…
AI For Mankind sees the potential of Deep Learning in early wildfire smoke detection and believes the advancements of computer vision will accelerate the adoption and deployment of vision based wildfire smoke detection system. …
We have been evaluating our wildfire smoke detection system against the real HPWREN cameras. This exercise gives us a deeper glimpse of various potential issues faced in real production setting. We would like to share some of our findings here.
We continue to curate, test, and share our results and datasets to encourage collaboration and speed up research/development in this area.
Fog vs Smoke
This video shows fog around the mountain top. Our model produced a false positive rate of 0.39 when tested against these images.
Our model did a good job in avoiding false positives during sunrise. The initial test produced a false positive rate of 0.01.
If you like what we are working on, please donate to support our efforts. All donations are tax deductible.
Each year, wildfires claim the lives of many people and cause significant environmental and economic damages. Early detection is key to effective fighting of wildfire. Once a wildfire reaches a certain size it can be hardly controlled.
Searching for More Images via Google Images/Bing Images
Data is the ingredient to build a machine learning model. For computer vision, images are the main training dataset. However, data collecting is a laborious process. In our Seeing The World project, we rely on volunteers to send us fruit/vegetable pics taken…