2021/05/13 by Shailesh Arya, Arya, Shailesh
Computer Science · Engineering · #Advanced Neural Network Applications #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Fire Detection and Safety Systems #IoT and Edge/Fog Computing #IoT-based Smart Home Systems #Machine Learning (cs.LG) #Video Surveillance and Tracking Methods
paper · pdf · doi:10.48550/arxiv.2105.06508
openalex publication_date 2021/05/13 · openalex created_date 2022/10/01 · openalex updated_date 2026/07/28
The vision of the internet of things (IoT) is a reality now. IoT devices are\ngetting cheaper, smaller. They are becoming more and more computationally and\nenergy-efficient. The global market of IoT-based video analytics has seen\nsignificant growth in recent years and it is expected to be a growing market\nsegment. For any IoT-based video analytics application, few key points\nrequired, such as cost-effectiveness, widespread use, flexible design, accurate\nscene detection, reusability of the framework. Video-based smart doorbell\nsystem is one such application domain for video analytics where many commercial\nofferings are available in the consumer market. However, such existing\nofferings are costly, monolithic, and proprietary. Also, there will be a\ntrade-off between accuracy and portability. To address the foreseen problems,\nI'm proposing a distributed framework for video analytics with a use case of a\nsmart doorbell system. The proposed framework uses AWS cloud services as a base\nplatform and to meet the price affordability constraint, the system was\nimplemented on affordable Raspberry Pi. The smart doorbell will be able to\nrecognize the known/unknown person with at most accuracy. The smart doorbell\nsystem is also having additional detection functionalities such as harmful\nweapon detection, noteworthy vehicle detection, animal/pet detection. An iOS\napplication is specifically developed for this implementation which can receive\nthe notification from the smart doorbell in real-time. Finally, the paper also\nmentions the classical approaches for video analytics, their feasibility in\nimplementing with this use-case, and comparative analysis in terms of accuracy\nand time required to detect an object in the frame is carried out. Results\nconclude that AWS cloud-based approach is worthy for this smart doorbell use\ncase.\n