Has Alibaba, which recently formed the “Dharma House,†also entered the world of autonomous driving?
On January 14th, a reporter from Ayao discovered that Alibaba made its debut on the KITTI road scene segmentation chart, one of the most prestigious benchmarks in automated driving computer vision algorithms worldwide.
The rankings revealed that Alibaba’s iDST AI team secured first place in three segmentation tasks, including two specialized scenarios—UU_ROAD (Urban Lane) and UMM_ROAD (Multi-lane)—as well as overall evaluation for URBAN_ROAD (urban lane).
Alibaba confirmed the news to Xinhua News but clarified that it is not entering the autopilot business directly. Instead, the company emphasized its focus on fundamental scientific research, with autonomous driving technology being a key part of that effort. The research is led by Ren Xiaofeng, an Alibaba machine vision scientist.
It is worth noting that the KITTI dataset serves as a crucial platform for autonomous driving development. Companies like Toyota, NVIDIA, Samsung, and NEC have all participated in its evaluations.
Established jointly by the Karlsruhe Institute of Technology in Germany and the Toyota Institute of Technology in the U.S., KITTI is widely recognized as the gold standard for evaluating autonomous driving algorithms. It covers five main scenarios: urban areas, residential zones, roads, campuses, and pedestrian environments. The bird’s-eye view (BEV) perspective, which is highly relevant to autonomous systems, includes complex road features that are challenging to identify.
According to the KITTI official website, the dataset is divided into categories such as scene classification, object detection, and tracking. A researcher told Xinhua News that road scene segmentation plays a critical role in autonomous driving. It is essential for defining drivable areas, path planning, high-precision mapping, and AR-based navigation systems.
Ali won the top spot in three segmentation tasks. Source: KITTI official website screenshot
While many autonomous driving teams can segment large urban roads, identifying edge cases—such as potholes, parked cars, or uneven surfaces—remains a challenge. Precisely segmenting the road's boundaries is a major focus in current research.
According to Alibaba, the iDST visual technology team, led by Ren Xiaofeng, developed an online learning approach using difficult sample mining. They also introduced an online data augmentation module, along with global and local feature integration mechanisms, to improve the network’s ability to represent road features. This helped address issues like inaccurate road edge segmentation and interference from shadows.
After six months of research, Alibaba achieved a remarkable 96.06% accuracy for the U.S. lane scenario and 97.70% for the multi-lane task. It also scored 96.76% in the overall road scene evaluation for URBAN_ROAD, securing the top position. This marked Alibaba’s first appearance on the KITTI road scene segmentation leaderboard.
In June 2017, Handan News reported on Ren Xiaofeng, the highest-ranked Chinese scientist at the AiDST team. Previously, he led the algorithm development for Amazon Go’s unmanned store. Upon joining Alibaba, he was praised by experts for his comprehensive understanding of computer vision, covering everything from image classification to event detection.
The reporter also found that Alibaba’s iDST team appeared in several other KITTI projects, including vehicle detection, pedestrian detection, and re-identification. These technologies have already been applied in Alibaba Cloud’s ET City Brain, used for urban traffic management.
Alibaba is continuously enhancing its scientific capabilities and investing heavily in basic research. At the 2017 Yunqi Conference in Yunnan, Alibaba announced the establishment of the “Dharma House†lab, aiming to explore future advancements in science and technology. The initiative plans to invest over RMB 100 billion in new technologies over the next three years, focusing on both foundational research and disruptive innovation. Jack Ma, Alibaba’s chairman, hopes that the Dharma House will become one of the best research institutions in the world.
Timing Belt,Lift Door Machine Belt,Elevator Door Machine Belt,Elevator Timing Belt
Shanghai Janetec Electric Co., Ltd. , https://www.janetecelectric.com