“In addition to cameras and sensors, artificial intelligence (AI) is another technology that merits rapid adoption for fire prevention, detection, and suppression.” – Independent Institute1
Fast & Accurate Dual-Sensor
Our LWIR detection is fast (as small as 2 m2 flame at 5 km).
Accurately detect fire occurrence by dual-detection (visual & infrared)
Our patented technology accurately determines fire location.
Work reliably in tough environments (dark, light polluted, hazy, windy & hilly).
Automatic & Efficient AI Recognition
Always-on Automatic AI detection. No delay nor tiredness.
Self-improving machine learning system analyses raw images on the edge
Large coverage (70k Ha per robot) and minimal manual operation.
More manpower-efficient, energy-efficient, and carbon-friendly than watchtowers and aerial fire patrol.
Intuitive & Versatile Management Platform
Easy-to-use GUI with 2D/3D maps. Supports cross platforms.
Situational awareness with real-time video & panoramic image.
Support visualization of 3rd party data via standard API.
Timestamped GIS-referenced fire records for post-event forensic and risk mitigation planning.
Figure 1: InsightFD Key Customer Benefits
In general, the ground-based wildfire detection system can detect ignitions more quickly than satellite and drone-based detections. However, most CCTV systems are expensive to deploy and operate:
- Expensive high capacity network to transmit HD video from expensive HD camera to the operation room
- They require additional manpower to operate the system. These operators are trained to use the proprietary system to identify wildfire smoke.
- They need highly focused manpower resources to fight false alarms. Visual fire detection is subjected to distractions: cloud movement, cloud shadows, chimney smoke, tree vibrations, and sunlight reflection of bright surfaces. It is a mentally demanding task for operators.
- Last but not least, visual fire detection is half-baked! Smoke is more difficult to spot at night. Civilians are sleeping at night and cannot help reporting smoke sightings. Firefighting agencies can only rely on ineffective visual fire detection!
Figure 2: InsightFD hardware comprises of a dual-sensor robot and edge computer
Automatic & Efficient
The purpose-built Dual-Sensor system in InsightFD captures infrared and visual images of flame and smoke respectively at the early ignition stage and achieve Fast and Accurate wildfire/forest fire/wildland fire/bushfire detection.
What happens if there is no line-of-sight and we cannot observe infrared emitted from the heat source? What if we can only see smoke behind the mountain? Is it a real forest fire smoke or some moving pine trees 15km away?
Different from traditional visual fire detection, InsightFD uses AI to process images. Our Machine Learning algorithms analyze the shape, color, motion, and patterns of smoke. We have been working with our customers to develop our image library since 2011. Our image library consists of hundreds of thousands of unique smoke images collected in the past nine years. We use large data sets to train our machine learning algorithms. They can identify smoke as good as experienced human operators! Our AI image recognition is automatic, objective, and untiring.
“By using machine learning techniques, you can build computer systems and applications that do tasks that are commonly associated with human intelligence. These tasks include image recognition, speech recognition, and language translation.” – Microsoft2
Figure 3: Deep Learning, Machine Learning, and AI. Source: Microsoft2
To achieve high precision visual detection, we need high-definition image input. Each InsightFD dual-sensor robot is coupled with a powerful industrial computer AI-Engine. Off-the-shelf cameras provide lossy compressed images. Our home-grown dual-sensor system provides uncompressed raw HD images to AI-Engine that running our machine learning detections.
Below are the unique customer benefits of our AI-based Detections:
- Automatic detection: Looking for fire and smoke around-the-clock consistently without hesitation, delay, and tiredness.
- Self-Improving Machine Learning System: Taking multiple inputs simultaneously to identify smoke. High true positive rates reduce unnecessary manpower resources wasted on false alarms. Online Learning improves our detection algorithms every time they process new inputs. The climate, landscape, and forest keep changing. We embrace the changes.
- Manpower-Efficient: No additional operators are required to operate InsightFD
- Cost-Efficient: Large detection coverage. InsightFD can detect flame and smoke at a distance of 15 km. A single robot can monitor an area of 70,000 Ha. A team of robots can cover a large wildland with hundreds of thousands of hectares.
- Network-Efficient: Each robot is equipped with an AI-Engine to do edge-computing. Large HD images are analyzed on the edge device. Data traffic between the robot and the Insight Globe management platform is small by sending just the detection alert information. InsightFD does not require high capacity network infrastructure.
- Energy-Efficient: each robot and AI-Engine pair consumes only 40-50W. InsightFD is more carbon-friendly than watchtowers and aerial fire patrols.
Figure 4: Online Learning. Source: Hands-On Machine Learning with Scikit-Learn and TensorFlow
“Information and warning dissemination delays due to human factor of hesitating to commit. We need a consistent, streamline process that mitigates delays due to the human factor.”3
We will discuss the Intuitive and Versatile aspects of InsightFD in the next blog, the final episode of this 3-part series. Please download our InsightFD brochure if you want to learn more about our product.
Reference:
- California Wildfire – Key Recommendations to Prevent Future Disasters, Independent Institute
- Deep learning vs machine learning, Microsoft Azure
- WUI Operational Requirements and Capability Analysis – Report of Findings, Department of Homeland Security, USA
