Innovations are being made to increase building energy efficiency as part of a key strategy to reduce carbon emissions.

The webinar series hosted by e4ds Make was released on the 7th at e4ds EEWebinar. Choi Ha-neul, a senior researcher at the Korea Institute of Energy Research, gave a presentation titled ‘Introduction to Building Control Research Using Computer Vision Technology.’

The final energy consumption in the building sector in Korea consumes 22% of the country's total energy consumption. It is predicted that the introduction of computer vision technology will enable a reduction of 5-10% in building energy consumption.

△Computer vision, which has high utilization in healthcare, facial recognition, autonomous driving, etc., can be built in cloud computing through system connection using only existing CCTV infrastructure and inexpensive camera devices.

Senior Researcher Choi introduced computer vision-based occupant-centric air conditioning control technology. The main actors in building energy are occupants, whose behavior can increase energy consumption by 80% or reduce it by 50%.

Accordingly, the concept of occupant-centered control is gaining attention, and the goal is to realize 'computer vision-based occupant-centered air conditioning control' that senses information about occupants in a building and controls the building.

Computer vision can detect and estimate the number of occupants through images, and can even estimate the amount of metabolism and clothing (the degree of clothing worn) by looking at the actions and clothing of the occupants. The air conditioning control will control the temperature based on PMV (Predicted Mean Vote) that reflects the amount of metabolism and clothing.

PMV, which stands for expected average thermal sensation or comfort, can be calculated by inputting temperature, relative humidity, metabolic rate, etc. Based on PMV, the optimal set temperature can be determined and the temperature of the air conditioner can be controlled, thereby improving the efficiency of the building's power consumption and implementing a green building.

Future challenges will be to resolve privacy and security issues in computer vision-cloud computing systems. Accordingly, the Korea Institute of Energy Research is said to be researching computer vision in edge devices.

Researcher Choi added, “Because the model for processing images is large, there are limitations to performing calculations at the edge, so we are working to overcome this issue.”