Using Shengteng AI technology, the Qinling·Qinchuan transportation model helps Xi'an build a smart transportation innovation center

"High complexity, high fragmentation, and cross-domain" have always been the primary pain points on the road to digital and intelligent upgrading of the transportation industry. Recently, the "Qinling·Qinchuan Traffic Model" with a parameter scale of 100 billion, jointly built by China Vision, Xi'an Yanta District Government, and Xi'an Future Artificial Intelligence Computing Center, is oriented to the field of smart transportation and provides services to Xi'an and its surrounding areas. The region will create a fulcrum for smart transportation innovation.

"Qinling·Qinchuan Traffic Large Model" combines Xi'an's massive local traffic ecological data in open scenarios, the original advanced algorithm independently developed by China Science Vision and the powerful computing power of Xi'an Future Artificial Intelligence Computing Center Shengteng AI , bringing digital transformation to all smart transportation scenarios such as road network monitoring, emergency command, maintenance management, and public travel.

Traffic management has different characteristics in different cities. There are great differences in the traffic characteristics, driving patterns and traffic flow of different roads. Therefore, targeted improvement measures are still needed in terms of traffic emergency management, special governance and comprehensive operations

In the field of traffic safety management, we face some challenges, such as the difficulty of fortifying illegal vehicles across provinces, the difficulty of comprehensive road safety management, and the problems of comprehensive law enforcement of operating vehicles such as garbage trucks and other specific scenarios. In order to deal with these pain points, the "Qinling·Qinchuan Traffic Model" uses its unique model capabilities to provide advantageous solutions

In the comprehensive management plan for special categories of vehicles, such as two-passenger and one-hazardous vehicles, garbage trucks, hazardous chemical trucks, etc., the "reuse" method of using cameras may be affected by light, shooting angle, image background and occlusion and other practical problems, thus affecting the identification and tracking of these operating vehicles. This deformable model structure can adaptively segment images according to the position and scale of the target, effectively overcome the interference factors caused by variables such as vehicle posture, angle and occlusion, and accurately identify various special categories of vehicles, end-to-end More reliably solve application implementation problems in end-end solution delivery

In the road management and maintenance scenario, the long lead time for PCI and PQI report delivery is caused by various factors such as multiple types of road disease data collection vehicles, complicated road surface disease data types, and difficulty in identifying millimeter-level diseases such as small cracks. , often rework problems, "Qinling·Qinchuan Traffic Model" relies on its excellent capabilities to assist highway owners such as expressways, national and provincial highways and other highways to effectively solve common pain points in the industry.

For example, in the identification of fine cracks on highways, the model has learned rich prior knowledge in the pre-training stage, so when it is transferred to a long-tail data set, it can more easily learn the tail categories. Feature representation. In practical applications such as the identification of small types of road diseases, its identification accuracy is also more accurate and efficient than traditional manual and semi-automatic marking methods. In the urban traffic congestion management scenario, the solution also needs to be combined with the city's traffic characteristics to assist urban traffic managers as a whole to

improve the comprehensive capabilities of traffic emergency management and situation analysis

. The "Qinling·Qinchuan Traffic Model" uses its own multi-modal sensing signal fusion capabilities and intersection dispatching decision-making capabilities to combine the urban road distribution of Xi'an's "chessboard road network, one city with multiple centers",

It can sense the status of vehicles, pedestrians, and roads at core intersections and trunk lines in real time. At the same time, based on the characteristic distribution of historical traffic data, it can assist urban traffic managers to provide corresponding traffic flow coordination plans, while reducing traffic congestion. This greatly reduces the possibility of accidents caused by congestion.

Going deep into the core business scenarios, the "Qinling·Qinchuan Traffic Model" currently has mature application capabilities such as comprehensive traffic safety management for urban traffic and intelligent data analysis and judgment. Its degree of intelligence is far superior to the current industry application status. .

In the future, China Vision will continue to cooperate with the Xi'an Municipal Government and the Xi'an Future Artificial Intelligence Computing Center to jointly explore the possibility of applying the "Qinling·Qinchuan Traffic Model" in aspects such as autonomous driving assistance and vehicle-road collaboration V2X. , based on Shengteng AI basic software and hardware platform. We will use our outstanding scientific and technological strength and digital innovation technology to empower urban transportation, promote the digital upgrading of urban transportation, and contribute wisdom and technological strength to building a transportation powerhouse!

The content about Zhongke Vision Language needs to be rewritten

China Vision is a company under the Institute of Automation, Chinese Academy of Sciences. It has focused on the field of artificial intelligence for more than 20 years and has accumulated rich technical experience. We are committed to developing core technologies for multi-heterogeneous sensing and fine-grained recognition, and provide a full range of artificial intelligence software and hardware platforms and solutions for the field of smart transportation

As a representative growing enterprise in the field of smart transportation, Sinovision has deeply explored the pain points of the industry based on years of deep cultivation and accumulation in the field of smart transportation, and is committed to using excellent technology to continue to explore the "optimal solution of the universal visual model in the transportation industry" ".

In the field of smart transportation, Zhongke Vision is currently committed to providing digital intelligent services for four transportation application scenarios: cities, rural areas, highways and transportation hubs. We use cutting-edge technology products to achieve this

About Xi’an Future Artificial Intelligence Computing Center

Xi'an Future Artificial Intelligence Computing Center, as the country's new generation of artificial intelligence public computing power open innovation platform, is the first large-scale artificial intelligence computing power cluster in Northwest China, and is based on "independently innovative artificial intelligence software and hardware infrastructure" as the key support. Focusing on building a center and creating four ecological platforms, it provides accurate and reliable model training and inference services to artificial intelligence universities, enterprises and scientific research institutes to meet the diversified needs of industrial development for computing power and realize the "policy-industry-industry" of artificial intelligence. A closed loop of “learning-research-application”.

Please slide the picture below

Learn more[Huawei China Government and Enterprise Business]Latest News

I can see you "looking" every time!

The above is the detailed content of Using Shengteng AI technology, the Qinling·Qinchuan transportation model helps Xi'an build a smart transportation innovation center. For more information, please follow other related articles on the PHP Chinese website!