Risk Management
Fengtu Technology needed to run GPU-heavy inspection and safety models in the cloud without wasting compute, while also improving fatigue detection and commercial-vehicle risk response for logistics customers.
Reported impact
Fengtu improved fatigue-risk detection, cut invalid alerts and operating cost, and turned GPU-heavy inference into an elastic cloud service with higher business reliability.
Fengtu Technology used Huawei Cloud FunctionGraph to build a serverless inference platform for transport safety and vehicle-quality inspection workloads. The platform turns model execution into event-driven functions that automatically scale GPU-backed instances with request concurrency, so the team can run inference and device self-check logic without managing fixed clusters. For commercial-vehicle safety, Fengtu combines driving-assistance hardware, a transport supervision platform, 24/7 operations support, and a cloud safety platform for risk prediction, real-time intervention, investigation reporting, and alarm workflows. Its DMS fatigue-detection pipeline uses cloud compensation and alert cleaning to detect fatigue, distraction, smoking, phone use, and camera occlusion, while a lightweight ResNet classifier checks camera status. Image-heavy self-test steps run inside FunctionGraph using fractional GPU allocation and built-in monitoring logs.
Images and driving-event inputs are routed into FunctionGraph, where GPU-backed functions run inference and self-check steps on demand. Detection outputs flow into the safety cloud platform, which cleans alerts, triggers alarms, supports investigations, and feeds risk-intervention workflows for commercial-vehicle operators.
"Fengtu built a three-in-one safety-driving solution with smart driving-assistance hardware, a smart traffic supervision platform, and 24/7 operational support. (original: 丰图科技还通过驾驶辅助智能硬件、智慧交通监督平台、7*24小时运营支撑,实现了三位一体的安全驾驶解决方案。)"
"FunctionGraph cut cost by 30% and raised reliability to 100%. (original: 华为云FunctionGraph助力丰图科技成本降低30%,可靠性提升100%。)"
"The DMS fatigue-detection pipeline automatically detects fatigue, distraction, smoking, phone use, and camera occlusion; camera-state accuracy reached 99%, invalid alerts dropped by more than 40%, and high-risk recall improved by 32%. (original: 针对长途货运司机容易疲劳驾驶的问题,丰图科技的安全驾驶解决方案引入了DMS疲劳检测... 基于精简版ResNet图像分类,对摄像头状态判断准确率达到99%。能够减少40%以上的无效告警和10%左右的护航人力投入,高危等级风险事件召回率提升32%。)"
"The platform uses FunctionGraph to make inference algorithms event-driven and automatically scale compute instances by request concurrency; GPU can be configured down to 1/16 granularity and billed by usage. (original: 该平台利用华为云函数工作流FunctionGraph，实现了推理算法事件驱动，全自动按请求并发量弹性计算实例，保障了资源及时响应业务请求；同时GPU最小1/16粒度配置,按使用量计费，避免算力闲置浪费。)"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Huawei Cloud Sample Archived · Dec 14, 2024 Shengtong Technology Shengtong Technology needed a way to make industrial inspection models more accurate and easier to adapt across power, mining, and chemical sites without buying large amounts of local GPU infrastructure. 3 billion parameters in the parent model Pangu mining foundation model · Shengtong industrial inspection large model · industry large models 
OpenAI Sample Oct 14, 2026 SentinelOne SentinelOne needed to reverse engineer sophisticated malware far faster than specialist researchers could do manually so it could investigate more samples, spot relationships across threats, and extend scarce cyber… 8 hours to reverse-engineer malware, down from 4–6 weeks OpenAI Daybreak · OpenAI models 
Google Cloud Sample Archived · Sep 25, 2026 Aldecis Aldecis needed to automate bottom-up financial analysis because CFO and FP&A teams were spending days hunting through dashboards to find the operational causes of margin and profit drift. 85% faster financial data analysis with BigQuery · BigQuery ML · Cloud Run 