1. Introduction:Today, in this era of widespread online social networking, there are various scenarios in which many accounts exist that are not what they claim to be or not what they represent to other users of the social network. The phrase "fake accounts" is now a widely used term in social networks around the world. In online social networks (OSN) the term fake account is associated with the term called Sybil attack. Sybil (means fake) attacks refer to the creation of fake nodes in a network intended for the purpose of dangerous or suspicious activities. The fake nodes examined in online social networks refer to fake users, each called Sybil, created to spread malware or spam throughout the network[. Sybil Attack has largely compromised the security of many OSNs including Facebook, Renren, Twitter, Tuenti etc. They were responsible for data leaks and access to private data or resources. It can also be responsible for "Clickfraud" when used to control and increase the number of views/visits on a single page. Sybil accounts are usually under the control of real but malicious users. They aim to increase their influence on the network by spreading and conquering the network. Therefore online social networks must be protected from Sybil attacks to prevent their accounts from being imitated or taken over by Sybil nodes. This mechanism is known as Sybil defense and uses various mechanisms to protect the network from Sybil attacks. Sybil defense mechanisms can only be incorporated once the network can detect the presence of Sybil accounts, so we need different Sybil detection techniques that take different approaches to detect the pr...... middle of paper...... owdsourcing: Name different categories of workforce (real humans) to detect Sybil accounts in social networking sites. The paper-based survey shows that real humans are better than default algorithms when it comes to detecting Sybil accounts as they easily adapt to changing account properties. The survey is mainly conducted on 3 major social networking sites, namely Renren in China, Facebook in the United States and Facebook in India. Data is collected from these sites and tested on three different categories of workforce, one the experts, two the Turkers and three the sociology university students, and the results are summarized and analyzed for their accuracy, dependency on demographics and various others factors. The above analysis is taken into account and a new practical crowdsourced Sybil detection system is designed which is more accurate and scalable