AI-powered IoT cameras are on track to add over 4 trillion kilograms in annual carbon dioxide emissions (Kg CO2e) by 2030, the equivalent of adding 860 million cars to the road in a decade.
Better AI has the potential to reduce these emissions significantly.
1Better AI can save 98.8% of carbon emissions than regular AI-powered IoT cameras.
Edge devices consume substantially less power than cloud devices. Here’s what we found:
An Edge device with a camera produces 4kg of CO2 per year.
* We assume the use-case requires near real-time framerate. Since we can already achieve ~24 FPS on a single RPi3B+ core, the max capacity per RPi3B+ is 4 near real-time cameras since it has 4 cores.
Network Access produces 123kg CO2 per year.
Cloud Inference produces 168kg CO2 per year.
* We assume that to achieve the same near real-time frame-rate (~24 FPS) per camera as the reference RPi3B+, a cloud GPU (without Xailient’s DNN) would only manage 5 cameras using YOLOv3 (since a YOLOv3 inference on a Titan X GPU (much faster than K80) is only 34 ms/frame on average).
The number of IoT devices is expected to reach 125-500 Billion by 2030, and assuming that 20% of them will have cameras, IoT devices with cameras are a 13-100 billion unit market. Considering that 12% of the market has Xailient Edge AI, 500 million tonnes of CO2 will be saved annually by 2030.
Each AI Camera device produces 4 KgCO2e per year. With Cloud AI, the network for data transmission produces 123 Kg of CO2 per year, and Cloud inference produces 198 Kg CO2e per year. In contrast, with Edge AI, networking and the Cloud are not required as processing takes place at the Edge device, closer to the data source, thus saving 98.8% of CO2 production per year.
Xailient saves 321 KgCO2e per AI camera device per year. A US car produces about 60,000 KgCO2e in its lifetime and produces 4600 Kg CO2e per year. With 15 Xailient AI installs, Xailient can save one car equivalent of CO2.
With 1.6 billion cameras, Xailient can save 500 million Tonnes of CO2e/year. The equivalent of taking 108 million cars off the road
Xailient Computer Vision systems are so efficient they can run at the Edge, even on existing hardware. When deploying new hardware, the Detectum software allows for smaller chips. Less computation and smaller hardware. This means AI with a smaller carbon footprint, and Edge deployment means less waste in transmitting data.
March 14, 2022
Xailient specializes in extremely efficient low-power computer vision. Intel's OpenVINO specializes in maximizing the performance and speed of computer vision AI workloads. OpenVINO improved Xailient FPS 9.5x on Intel hardware to 448 FPS. Together, Xailient-Intel outperforms the comparable MobileNet_SSD by 80x. Even after Intel worked the OpenVINO magic on MobileNet_SSD, Xailient-OpenVINO is 14x faster.
November 29, 2021
Xailient’s Face Recognition enables high-speed edge AI processing with low-power consumption using Sony’s IMX500 – a chip so small it can fit on the tip of your finger.
We see things differently in the dynamic field of computer vision AI
You’ll get insights and resources into: