#Aivon #aivonico #tokensale #AI #Blockchain #aivonio
Brief Introduction
AIVON is decentralized block-based system and protocol built on the AI network and a community of human professionals working jointly to produce normalized and enhanced metadata for video data.
Uses of Artificial Intelligence
Expertise AI algorithms will be deployed on Artificial Intelligence nodes, of which the GPU and CPU resources can be applied to scan media files, produce the improved metadata including classification, time-coded tags, transcripts, categories and translations, and an index of the video content. Each one with reasonably powerful PC can establish an AI node and contribute the AIVON protocol. The load will be decentralized and allocated across the AIVON protocol, same to how SETI HOME and the many other distributed computing plans time-share computing recourses.
Every video content, such as a full program or extracted part, will be assigned and sole identifier using the EIDR or entertainment Identifier Registry. This gives worldwide unique identifiers for the complete range of audiovisual content types that are related to entertainment commerce. AIVON protocol nodes will be saved their metadata in JSON module anchored to the ETH block. Most of the metadata will be saved directly in the block system but connected to it applying EIDR identifier. This is just because the value of the data can be quite huge, particularly when you add translations and transcripts.
A main feature of this plan is the generation of trade mark content securely index known as a content graph. The AIVON will use AI to define a confidence rate of each of many content security attributes, such as offensive language, guns, alcohol, hate speech, illegal drugs, adult, violence, nudity, religion etc. these confidence rates 11 will be combined into an array of figures. This short string of figures, named a content graph, will let programmatic matching of content as well as context, for optimal.
A content graph can be visualized as a bar graph where video content attributes are plotted along the vertical axis and confidence scores plotted on the horizontal axis. There'll be attributes where the confidence NIL and function is not detected. The left to right ordering is reserved backed on the schema so that 2 or more content graph patterns can be compared.
Accessing attributes engages a solo subscript which represents a range index. The range of positions corresponding to exact attributes being recorded can boost over time, without affecting any other uses of the content graph.
Is AI any fine at detecting scenes involving violence nudity or guns?
Yes to some degree. It has been shown that AI algorithm can 95% perfectly detect video objects such as guns within content - even rough CCTV footage.
Detection of Nudity is a little more challenging, but it has been reported to be dependable on still pictures. According to single widely referenced document on the subject: “the AI algorithm is indeed capable to provide fine recognition prices for nudity even at the frame pointer, achieving, in the finest case, a value of 93.2 percent of right classification. The outcomes also point to that the proposed voting plan significantly improves the recognition prices for video content segments, solving a few ambiguities and frames misclassifications”. This shows that nakedness detection involving a temporal part is even more correct.
D App and conventional apps can use the Content Graph to mechanize the determinations of content correctness and brand security. They believe that the Content Graph will not just solve several problems for the video content field, such as search and discovery, but it can also turn out to be an industry standard much like MPAA Rating.
For more information:
• Website: https://aivon.io/
• Twitter: http://www.twitter.com/aivonio
• Facebook: http://www.facebook.com/aivonio
• Telegram: http://t.me/aivonio
• Whitepaper: https://aivon.io/download-whitepaper/
my bitcointalk profile - https://bitcointalk.org/index.php?action=profile;u=1352940;sa=summary
ETH address - 0x41253E34A9D53B1eD0a2c53e5418B607eDf6A301