A Study to Contribute to Artificial Intelligence Training from Meta

Meta has officially launched Casual Conversations V2, a new database to be used in training AI models. Thanks to this data bank, many individuals that were not recognized correctly by artificial intelligence models will be perceived more successfully.
 A Study to Contribute to Artificial Intelligence Training from Meta
READING NOW A Study to Contribute to Artificial Intelligence Training from Meta

With the rapid development of artificial intelligence, artificial intelligence systems have started to be used in all areas of our lives. In addition to artificial intelligence models such as ChatGPT and DALL-E, artificial intelligence-supported filters and facial recognition systems have also become popular in Snapchat and TikTok.

However, at present, artificial intelligence has difficulties in recognizing and perceiving individuals of different gender, race, physical appearance and disabilities. Meta, on the other hand, organized a new research to find a permanent solution to this problem.

More than 26,000 videos have been recorded to be used in training artificial intelligence systems

In addition to voice artificial intelligence systems like Siri and Google Assistant, Meta, which created a database called Casual Conversations (CC) in 2021 for chatbots, systems using facial recognition and visual artificial intelligence models, contributed to the development of many artificial intelligence models. The company aims to enable these systems to detect much more individuals with the second generation CC database that it has prepared in the past months.

In this data bank, which consists of more than 26,000 videos recorded by 5,000 participants from 7 countries, there are also videos of individuals with different appearances and physical or mental disabilities.

According to Meta, second-generation CC-trained AI systems will be able to detect people with freckles, limp people, individuals with physical or mental disabilities, and all “out-of-the-box” individuals.

In addition, Anya Taylor-Joy and Colin Firth want to be able to distinguish between celebrities and “celebrity-like individuals” by taking samples from individuals who look or sound similar to celebrities in the database.

Meta also makes the Casual Conversations 2 database publicly available, stating that anyone who develops an AI model can train their algorithms with this database. In addition, anyone who wants to contribute to the data bank can help further development of artificial intelligence by shooting videos.

Comments
Leave a Comment

Details
136 read
okunma56167