news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

By Li Anqi

Editor: Li Qin

36Kr learned that recently, AI data company "Integer Intelligence" announced that it had completed a round A financing of tens of millions of RMB. This round of investment was led by Frees Capital, and old shareholder Oufang Angel followed suit. Previously, Integer Intelligence also received investments from Tang's (China) Investment, Spinel Capital and other institutions.

It is reported that the funds from this round of financing will be used to strengthen the construction of data set construction services and intelligent data engineering platforms, recruit the world's top talent team, and accelerate the company's product and business layout in the global market.

Integer Intelligence is a data company dedicated to providing data service support for artificial intelligence. Its business covers many fields such as intelligent driving (Automobile AI), generative AI (Generative AI), embodied AI, etc. The current team size of the company is less than 100 people.

The development of artificial intelligence is inseparable from the three pillars of data, algorithms and computing power. With the accelerated explosion of AI big models, the demand for data is growing exponentially, and high-quality data is crucial to improving the accuracy and generalization ability of AI models. Integer Intelligence is committed to significantly reducing the cost and time of data preparation through effective data management and intelligent labeling tools.

According to the company's founder Lin Qunshu, Integer Intelligence's data services have two major characteristics: one is that they cover all business scenarios, and the other is that they provide a data platform that can be automated with AI and continuously evolve and upgrade.

First, the data set construction service (ACE Service) provided by Integer Intelligence covers all aspects such as data collection, cleaning, labeling and finished data, providing solutions for the fields of intelligent driving, generative artificial intelligence, embodied intelligence, etc. In addition, Integer Intelligence has also formed a training system for artificial intelligence trainers and established multiple artificial intelligence data labeling industry bases around the world.

Taking the intelligent driving scenario as an example, high-level intelligent driving has become a smart technology highland that car companies must compete for, and the acquisition and processing of massive intelligent driving data has also become a challenge.

Image source: official

For the data processing industry, the original 2D3D data annotation tool chain can no longer meet more requirements. Therefore, Integer Intelligence has launched a 4D annotation tool, which reconstructs the scene by superimposing sequence information of the time dimension on the basis of three-dimensional spatial data. This can compress the annotation work that originally required hundreds of frames to just one frame.

According to Lin Qunshu, Zhengcheng Intelligence launched the 4D labeling tool at the beginning of last year. With the optimization of the tool chain, it can not only achieve a hundredfold increase in efficiency, but also achieve an automatic labeling accuracy of up to 98%, winning orders from many leading domestic companies.

Generative AI and embodied intelligence are also business scenarios targeted by Integer Intelligence. For generative AI, Integer Intelligence provides data collection, data cleaning capabilities, and data annotation tools, and reserves a rich multimodal finished data set. It also accumulates expert library resources in the fields of mathematics and programming for the production of training data sets in professional fields; for embodied intelligence, Integer can also provide data sets containing multiple sensory data such as vision, touch, and sound.

Secondly, in terms of data platforms, the intelligent data engineering platform (MooreData Platform) of Integer Intelligence can process multimodal data such as images, videos, text, audio, point clouds, etc. With the support of AI intelligent tools, the efficiency of data annotation is improved by more than 500%. Through the platform's built-in MLOps model training system, the closed loop and iteration of data annotation engineering and model training engineering are realized, which can quickly perform AI intelligent annotation in new industries or new scenarios, and continuously improve the accuracy of AI intelligent annotation, so that the time cost and labor cost of obtaining high-quality data are continuously reduced.

Lin Qunshu believes that the wave of artificial intelligence and the growth of data labeling business will occur on a global scale, and the business focus of integer intelligence will also shift from focusing on the domestic market to a global layout.

At present, Chengxin has established cooperative relationships with more than 1,000 top technology companies and research institutions at home and abroad, set up Global Offices in Singapore, Paris, France, Silicon Valley, the United States, and established AI data trainer industrial bases in Southeast Asia and Africa.

After this round of financing, Lin Qunshu said that he will continue to strengthen the foundation of Integer Intelligence's 4D labeling business, advance the development of synthetic data, and promote overseas brand building and global business layout.

Regarding this round of financing, Frees Capital said that "lack of cutting-edge data" has become the biggest obstacle to the development of the AI ​​industry, and the solution is "data production" in a human-machine combination. Even top companies like OpenAI need to rely partly on external partners to achieve the goal of data production. Integer Intelligence is such an excellent "AI data factory". They started in China and have made good progress in overseas markets. With the support of multiple factors such as China's surging overseas wave, efficient digital foundation and engineer dividends, Integer Intelligence is expected to achieve greater breakthroughs and achievements in Europe, the United States and other international markets in the future.

Angel Oufang said that big models, embodied intelligence, and autonomous driving are leading a new revolution in artificial intelligence technology. High-quality labeled data directly affects the AI ​​model's ability to understand and learn specific tasks. The MooreData intelligent data labeling platform launched by Integer Intelligence can achieve low-cost, high-precision, and fast output of efficient and automated data labeling data production, thereby shortening the AI ​​model development cycle. As Integer Intelligence's global business continues to land, we believe that Integer Intelligence will become the infrastructure leader of the new generation of AI industry and contribute Chinese strength to the development of the world's AI industry.