Event detection in videos is becoming an emerging area of research now a day. Monitoring of people activities using a surveillance camera is an essential one in a recent lifestyle for safety and security. The surveillance cameras are used in a wide variety of places such as in public places, Hospitals, Schools, and Homes for the beneficiaries of common people, patients, children and the elderly. In case of any emergency or abnormal events, immediate notification should be given to the respective people. The abnormal events are recognized from the videos using deep architectures. The goal of event detection in videos is to detect simple and complex actions in real-time data. This has a lot of attention in real-time ambient assisted living environments especially for elder people who live alone in the home. In this paper, a deep architecture of long short term memory recurrent network is proposed to detect fall actions in video inputs..

Twitter being a famous social media site not only helps people to share their thoughts in microblogs but also plays a pivotal role in situations of emergency for communication, announcement and so on. However, it results in anaversive effect when inappropriate tweet is reposted or shared to people thereby spreading rumors. This work describesthe methodologies in identifying the rumors using specific attributes like precision, fi-score, recall and support thereby solving the ranging rumor issues across the social media platform. A system detects candidate's rumor from twitter and then evaluates it applicably. The result of experiment shows the proposed algorithm in order to detect the rumors with acceptable accuracy.