Artificial Intelligence Based Smart Robot Predicting Human Activities

• Published in: 2022 1st International Conference on Computational Science and Technology (ICCST) pp. 166 - 172
• Main Authors: Macriga, G. Adiline, D, Tharun Raj, J, Sneha Jaiswal
• Format: Conference Proceeding
• Language: English
• Published: IEEE 09.11.2022
• DOI: 10.1109/ICCST55948.2022.10040414

Perceiving human activities from video groupings or still pictures is an inciting task due to issues, for occasions, establishment wreck, midway hindrance, changes in scale, viewpoint, lighting, and appearance. Human action acknowledgment is a difficult time series order task. It includes anticipating the development of an individual in light of sensor information and generally includes profound area mastery and techniques from signal handling to accurately design highlights from the crude information to fit an Artificial Intelligence (AI) model. Human activity recognition assumes a critical part in human-to- human and human-computer association. Physically determined frameworks are exceptionally tedious and costlier. So, to plan a Human action recognition framework turns into a requirement for this present world situation. This framework won't just be practical yet in addition as a utility-based framework that can be consolidated in an enormous number of uses that will save time and help in different exercises that require acknowledgment cycle and save a great deal of time with great precision. The current Human Action Recognition (HAR) models are costlier that can't be managed for ongoing application. In this undertaking, video dataset is utilized for planning the framework. Highlight extraction strategies like optical stream and spatial fleeting procedures are being used to remove the elements. These datasets are trained using VGG19 after BoW Vector Extraction to predict normal and abnormal actions of human. The abnormal actions are live streamed to Police officials through a mobile application. In the interim, the doors of the automatic teller machine will be consequently shut to keep the robbery from taking off. Caution can be turned on to alarm the close by individuals. Subsequently, this task effectively gives a Human Action Recognition model which can be utilized progressively applications, for example, an answer for forestall and give proof of an Automatic teller machine, burglary.