BYD patents underbody vision system to detect life

CarNewsChina reports that Chinese automaker BYD filed a patent (application CN122200729A) for an underbody vision system that detects living organisms beneath parked vehicles. Per CarNewsChina, the application was published by China's National Intellectual Property Administration on June 12, 2026. The disclosed method captures a baseline image of the vehicle underside when the vehicle is powered down, stores that reference, then compares later real-time images to identify changed regions. CarNewsChina describes a two-step workflow: isolate delta regions, then run feature extraction and classification on those target areas to determine whether a living organism is present and assess its status, which the outlet says is intended to reduce false positives and lower compute needs.
What happened
CarNewsChina reports that Chinese automaker BYD has filed a patent application describing an underbody vision system for detecting living organisms under parked vehicles. Per CarNewsChina, the patent application is numbered CN122200729A and was published by China's National Intellectual Property Administration on June 12, 2026. CarNewsChina describes the filing as focused on a baseline-image comparison method that records the vehicle underside while the vehicle is powered down and later compares new images to that stored reference.
Technical details
Per the patent description reported by CarNewsChina, the workflow is two-stage. First the system captures a baseline image of the underbody and treats unchanged components (battery enclosures, suspension, aerodynamic panels) as static background. Later, real-time images are compared to the baseline and only regions that differ are isolated as target images. Second, the system performs feature extraction and applies detection algorithms to those target regions to determine whether a living organism (animal or person) is present and to evaluate its condition or status. The architecture is intended to focus compute on changed areas rather than processing the entire underbody scene each cycle.
Industry context
Industry observers note that a two-stage pipeline (delta or change detection followed by targeted classification) is a common pattern for edge vision systems when teams need to reduce false positives and conserve on-device compute. For vehicle safety use cases, underbody detection is technically harder than in-cabin monitoring because of variable lighting, shadows, debris, dirt accumulation, and uneven ground, which CarNewsChina highlights as challenges the patent seeks to address.
What to watch
Observers will look for whether the application proceeds to grant, whether similar patents appear from other OEMs or Tier-1 suppliers, and how manufacturers choose sensor modalities and compute partitioning for underbody checks (for example, visible-light imaging combined with thermal sensors or on-board accelerators). For practitioners, the filing illustrates a concrete example of delta-based processing that can be adapted for other constrained safety sensors where stable background reference frames are available.
Scoring Rationale
A notable patent filing in automotive computer vision showcasing practical delta-based detection for edge safety checks, relevant to practitioners building deployed vision systems. Scored as a patent application without production timeline, not a frontier research breakthrough or deployed technology.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems


