'Space' consists of an API that analyzes and classifies space. 'Taker' improves the understanding of space by analyzing each space and returns reports about it, and we are in the process of developing an API that can rapidly and accurately find a specific location within an indoor or outdoor map.
Taker is a Spatial Anlaysis AI service that analyze snapshots of specific spaces using computer vision and machine learning technologies. It can quickly analyze and recognize various objects located in the space and classify the recognized objects into four different spaces, such as living room / room / kitchen / bathroom. It is possible through the algorithm developed by Urbanbase. It can be used to recommend interior items, furniture, home appliances, and other products that fit well in to the space. It is being developed to the stage where persona analysis and customer preference analysis (age group, gender and personal preference) are possible.
With 80,000 space images and Generative Model (GAN), machine learning is done in multiple angles. It is about 96% accurate.
The object detection technology is based on the Faster-RCNN and will be extended with additional algorithms to increase both accuracy and speed.
By analyzing gender, age, residential type and personal style on the spatial image, users' preference can be predicted and used not only for e-commerce, but for various industries.