- Core Technology
- Based on the foundation of proprietary technologies and SenseCore AI infrastructure,
SenseTime has rapidly opened up AI application in multiple vertical scenarios, and is empowering various industries.
- 02Computing Platform
- 03Deep Learning Platform
- 04Model Library
01 / 05
SenseCore is the universal AI infrastructure created by SenseTime. Through integrating strong computing power and leading algorithm capabilities, it can re-discover the potential values of information, breaking the boundaries of AI knowledge and applications and reducing the production costs of AI applications, thereby achieving efficient, low-cost and scalable AI innovation and empowerment.
02 / 05
Powered by SenseTime’s proprietary new generation Artificial Intelligence Data Centre (AIDC), SenseCore integrates AI chips and AI sensors to provide a strong computing power foundation, supporting the analysis, training, and reasoning of massive data and algorithm models. AIDC is Asia’s largest artificial intelligence computing center located in Shanghai. It possesses a total computing power of 3,740 Petaflops and can complete the training of a 100-billion-parameter model within a day.
03 / 05
Deep Learning Platform
SenseTime creates an innovative AI universal algorithm development platform by integrating training data management platform, the proprietary deep learning training framework (SenseParrots), reasoning engine for deep learning (SensePPL) and a model factory, which provide comprehensive tools from training data storage and annotation to model training, production, deployment and testing, realizing large-scale production of performant neural network models. SenseParrots is seamlessly connected with the training data management platform and large-scale computing infrastructure and possesses a parallel processing efficiency of nearly 90%, achieving 900 times faster calculations in the neural network model training process. SensePPL is based on SenseTime’s proprietary computing base and reasoning engine, which provides excellent end-to-end reasoning ability.
04 / 05
The model library aggregates innovative trained neural network models, enabling businesses to leverage AI applications for intelligent upgrades. SenseTime plays an important role in the AI open-source ecosystem by partnering with The Chinese University of Hong Kong to offer OpenMMLab. SenseTime has accumulated over 34,000 neural network models. Its open-source algorithm framework has received over 50,000 stars on GitHub.
05 / 05
SenseCore enables efficient and scalable AI model production with minimal human and computing resources, which realizes automatic and self-adapted production and applications of AI algorithms, brings Smart Business, Smart City, Smart Life and Smart Auto to life, and boosts the industrialization of AI technology.
1.Accelerating large scale AI applications: SenseCore enables efficient and scalable AI applications, reducing the production cost of scenario-specific models
2.Addressing long-tail scenarios: SenseCore provides the capability of addressing long-tail scenarios in city management, enterprise services, people’s life, integrating AI into more value chains
3. Creating new business models: Digitalizing and analyzing the physical world to facilitate intelligent decision-making, forming new business models
- Computing Platform
- Deep Learning Platform
- Model Library