OBJECT ACTION DETECTION USING DEEP LEARNING

Object Action Detection is an intelligent video surveillance system developed to strengthen home security bydetecting abnormal human activities from recorded video footage. The system employs deep learning techniquesto analyzeActions and identify any suspicious behavior within the home environment.At the core of the system is a Spatial Autoencoder, which enables effective and accurate detection of unusualactivities in the video without high computational costs. Instead of relying on live streaming, the system processespre-recorded videos, detects abnormal actions, and captures key frames where such activity occurs.When a suspicious action is identified, the system automatically generates a mail alert to the homeowner, attachingthe relevant frame image for immediate review. These captured images are also stored for future reference,providing a reliable and intelligent way to monitor security events within the home from recorded footage.