How Alibaba’s City Brain Uses AI to Transform Urban Management

City Brain Overview At the recent Cloud Xi conference in Nanjing, Hua Xiansheng, deputy director of Alibaba’s DAMO Academy Machine Intelligence Lab, delivered a keynote titled “City Digital Transformation – City Brain,” introducing three AI products and sharing three future considerations for the City Brain.

The core challenge of urban perception is threefold: fragmented sensing ("blind men feeling an elephant"), limited intelligent camera capabilities, and an inability to uncover the causes behind events ("seeing flowers in the fog"). Video data, as pixel‑level information, offers comprehensive city‑wide perception that has been underutilized.

AI‑Driven Perception and Decision‑Making By analyzing video streams to detect people, vehicles, and events, the system enables real‑time alerts for accidents or congestion, allowing rapid response and resource allocation. Continuous perception also supports searchable digital information for tasks such as tracking hit‑and‑run vehicles or locating missing persons.

Predictive and Interventional Capabilities Leveraging current traffic flow and multimodal data, the platform predicts short‑term and longer‑term traffic conditions, enabling proactive interventions such as reallocating police resources or controlling traffic flow to prevent congestion.

Three AI Products

Tianyao provides real‑time, city‑wide detection and automatic patrol alerts for traffic events and accidents, turning non‑intelligent cameras into intelligent sensors using cloud computing and deep visual analysis. It can detect spatial and temporal anomalies, such as unusually high alarm frequencies at specific locations, leading to targeted traffic improvements.

Tianying achieves a world‑leading pedestrian recognition accuracy (>96%). It processes massive city video data and enables progressive video search, allowing rapid location of specific objects such as missing persons or hit‑and‑run vehicles. The underlying technology won first place in the KITTI benchmark and the Market‑1501 re‑identification dataset.

Tianji is a predictive product that forecasts vehicle and pedestrian flow, providing short‑term traffic forecasts (e.g., ten‑minute traffic conditions) and longer‑term predictions based on weather and city events, enabling better navigation, control, and early warning.

Future Thoughts on City Brain

The core is to continuously leverage advancing AI and computing power to extract value from ever‑growing urban data.

It tackles problems beyond human capability due to the complexity and scale of calculations, data, and issues.

The system is envisioned as essential urban infrastructure, akin to water and electricity, becoming indispensable for city operations.

The City Brain platform, selected by the Ministry of Science and Technology as one of the first four AI open‑innovation platforms, includes a foundational computing layer, data intelligence, intelligent applications, and industry‑specific solutions across transportation, safety, and future domains such as environment, healthcare, communications, and industry.