How does Xiaomi EV's "Leading Vehicle Recognition" feature work?
Official answer
Xiaomi EV's "Leading Vehicle Recognition" feature is primarily based on the Xiaomi self-developed multimodal large model, which uses powerful image understanding capabilities to identify the brand, model, and other information of the vehicle ahead in real time. While driving, you simply need to wake up the in-car voice assistant "Xiao Ai" and ask, "What car is driving ahead?" to quickly get the recognition result. The implementation of this feature relies on three core technological advantages:
Precise Recognition and Intelligent Coverage: Through intelligent algorithms, the target vehicle is precisely locked. Combined with multi-angle dynamic analysis, it can adapt to different weather, lighting, and complex road conditions, achieving full-scenario coverage on highways, urban roads, and more.
Data Collection and Training: The R&D team collected 50 to 100 photos of key angles, such as the front, rear, and sides, for each car model, covering diverse scenarios like day and night, near and far, to achieve the recognition of different models.
Technology Fusion and Optimization: Our R&D team adopted multimodal data fusion and sensor fusion, pairing image and semantic annotation data, and fine-tuning the data to improve the model's understanding of orientation. Currently, the "Leading Vehicle Recognition" feature of Xiaomi EV has an accuracy rate of 90% for common models, and the accuracy rate for Xiaomi's own models has exceeded 99%.
This feature is now fully available on all Xiaomi EV models currently on sale, and you can experience it by waking up "Xiao Ai". It currently supports the recognition of over 1,000 common vehicles, covering various models available in the Chinese market. No extra operation is needed; you can experience this feature simply through voice interaction, adding practicality and fun to human-machine interaction.