“Mohsen Yazdinjad”, the artificial ielligence lead team and the manager of Nextra Studio startup, iroduced the details of the fifth edition of Facebook’s artificial ielligence eve at the second Pandora startup gathering. This eve, which is known as one of the most importa artificial ielligence eves in Iran, is held with a focus on recognizing tourist places and people’s gender through machine vision.
An overview of the previous courses of the Facebook eve
Iroducing this eve as one of the biggest specialized artificial ielligence eves in Iran, Mohsen Yazdinjad meioned the history of this eve. “This eve is now in its fifth year,” he explained. “Each year, the eve focuses on a specific issue in the field of artificial ielligence to give participas the opportunity to work on it in depth and come up with innovative solutions.”
In the last four editions, participating teams have faced various issues to participate in this eve. The fourth course was dedicated to the topic of “Deepfake detection”, the aim of which was to distinguish healthy videos from problematic videos through artificial ielligence tools. Yazdinjad emphasized that these experiences helped the teams to improve their skills in more specialized areas of artificial ielligence.
The focus of this year’s eve is the recognition of tourist places is
The main focus of this year’s eve is the recognition of tourist places through the use of artificial ielligence. Yazdinjad explained: “Participating teams must recognize differe places using tourism images. This project uses the combination of machine vision technology and computer vision to correctly ideify tourist places.
He added: “In addition, ideifying the gender of the people in the images is also part of this eve; In this way, the participas must specify the number of men and women in the picture.
Competition structure and evaluation process
Yazdinijad we on to explain the structure of the competition. He explained: “The competition data is divided io three differe parts: the training data (Train Data), the public test (Public Test), and the final test (Private Test). “Participas can first train their models using training data and prepare them for competition.”
The director of Nextra explained: “The public test is also updated in stages and participas can earn more pois during the eve by improving their models. “Finally, the final test, which is used to accurately evaluate the teams, will determine the final result.”
He claims: “One of the key pois in this eve is to preve fraud and ensure the correct implemeation of the evaluation process. For this reason, the top teams must eveually run their code in the eve environme to validate the results. This action is taken to create fairness and transparency in the evaluation of teams and to ensure that no team has reached the result by cheating.”
In another part of his speech, Yazdinjad meioned the connection between the top teams and the sponsors of the eve. “One of the importa goals of this eve is to ideify top tales and connect them to job opportunities and industrial projects,” he said. “In previous editions, many top teams were able to pitch their projects to eve sponsors and receive outsourcing coracts.”
Yazdinjad also emphasized that top teams can use these opportunities to develop their businesses or cooperate with big companies.
Finally, expressing optimism about the success of this year’s eve, Yazdinjad emphasized that the main goal of the eve is to develop artificial ielligence technologies and ideify and cultivate new tales in this field.
