In recent years, the creation of new solutions and products in the field of artificial intelligence has further increased interest in technological innovations. The startup of "Ainuverse" LLC which received a "Startup" certificate from the Agency for the Development of Small and Medium Business, has developed the software "ainuvision," which is also considered one of the notable products in the field of artificial intelligence.

The creator of project is Mr. Orkhan Hasan, a young Azerbaijani and the founder and CEO of LLC, a specialist in artificial intelligence and the author of the 8th-ranked facial recognition algorithm in the world, as per a competition held by the National Institute of Standards and Technology (NIST) of the U.S. Department of Commerce, involving more than 400 influential foreign companies. Mr. Orkhan Gasanli, providing information about the project, stated that the "AinuVision" software simplifies methods of managing and accessing video content using artificial intelligence technology, thereby saving time and resources. "AinuVision" comprises an artificial intelligence algorithm that enables image search based on text queries using “Computer Vision” and “Natural Language Processing” technologies.

The program analyzes processes and issues warnings by detecting human images in files of any video format or on surveillance cameras. Currently, the solutions for video content recognition or intelligent cameras worldwide typically rely on specially trained traditional AI algorithms and can recognize only a few events. "If we seek these solutions or cameras to recognize all events, we need to create new additional AI models for each specific category. 'Ainuvision,' with its artificial algorithm, easily recognizes many events that people encounter." The main goal is to eliminate human factors and errors in processing visual images, ensuring faster and more accurate execution of video search processes.

The startuper mentioned that currently, it takes approximately 30 minutes to watch a 4-hour video at 8x speed while searching for specific moments. Top of Form Bottom of Form This can lead to missing critical moments. The aim of the new software is to find any image in long video files or clips in real-time within seconds based on included content or provide real-time alerts: "For example, if someone steals a bicycle at night and is caught on camera, and you search it instantly."

If we program the software to search for someone who stole a bicycle, it will display the timestamp with that image from the camera. Alternatively, using the application, you can set up a system for preemptive alerts. For instance, you could set it “to alert you when a white car approaches your door”. In this case, the alert would come to you at any time a white car approaches your door. This is about live videos, i.e., related to cameras. But if you can write it to find the image you're looking for in the movie, someone is drinking tea in the glass, find it. In this case, artificial intelligence will find and detect all tea drinkers. But if you say "someone drinking tea on an airplane," it would simply find the person drinking tea on the airplane and show it. One of the features of this software is that it understands and analyzes images based on natural language input, without the need for specific keywords. The project author believes that the software's "smart" capabilities offer extensive application possibilities in areas such as ensuring urban safety, safeguarding people's security, and maintaining public order. 

It allows for finding images of specific events in shopping centers, restaurants, and service facilities to enhance the security of staff and customers, or to prevent incidents by triggering an alarm signal upon the occurrence of predefined warning events set in the software. The uniqueness of the software lies in the capability to search for the desired image within a video file or in real-time mode based on the content of text entered by the user. In modern Video Management Software (VMS), artificial intelligence models used in cameras are trained on specific categories. As a result, when the model encounters an image that it hasn't been trained on, the system fails to recognize elements related to this new category. Based on surveys and virtual meetings with major international competitive organizations, as well as conducted research, we can assert that currently no system utilizes “Natural Language Processing” and “Computer Vision technologies” for content-based image search. This creates an opportunity to collaborate with organizations that have a large customer base in the future, offering the "AinuVision" software and integrating it into existing systems.

Search systems used in the international market not only employ traditional AI models but also utilize high-energy-consuming equipment (~700 W). Sale of the "AinuVision" software alongside energy-efficient partner equipment (~20 W) will provide "AinuVision" LLC with a competitive advantage. According to the startuper, initially, the software was intended to be sold in Baku. Since the majority of target user segments organizations dealing with video content, government institutions, and companies using surveillance cameras are located in Baku, the project implementation will also be feasible in the capital. Following this, depending on the target client segment, software sales will be organized in Central Asian countries (Kazakhstan, Uzbekistan) and Southeast Asian countries (Indonesia, Thailand, Malaysia, Vietnam). Penetration into the Asian market will be carried out jointly with partner companies in this region. The project's author noted that there is consideration for selling the "ainuvision" software not only to the video surveillance market but also to organizations involved in video editing and processing: "According to estimates, the software will access the market in 3 stages. The first stage is its sale, during which the software will be sold either independently or along with the existing equipment already available on the market.

The core of the business model is to sell the software with an annual license. The second stage involves joint sales of the software to equipment manufacturers who are partners and for this purpose, relationships have already been established with a partner company in Singapore, and initial software testing has been conducted on their equipment. Its significant energy savings and low price compared to competitors on the market are major advantages of the equipment. The third stage involves selling the software to camera chip manufacturers. It is technically challenging to integrate artificial intelligence algorithms into existing camera chips. Currently, the partner company manufactures next-generation chips for cameras, and in the future, integrating the software into camera chips will provide access to larger markets”. https://vergiler.az/news/economy/30697.html

Powered by Froala Editor