Page 132 - AI for Good-Innovate for Impact Final Report 2024
P. 132

AI for Good-Innovate for Impact



                      Use Case – 29: AI-based Video Analysis System to Intelligently

                      Protect the Lives of Miners











                      Country: China

                      Organization Name: Shenzhen Corerain Technologies Co.Ltd.

                      Contact: Dai Jie, jie.dai@ corerain .com


                      29�1� Use case summary table

                       Domain               Mining, Protecting Lives

                       Problem to be        Protect lives of miners and safety of mining equipment
                       addressed

                       Key aspects of the   •  Streaming accelerator for AI
                       solution             •  Safety surveillance and monitoring system
                                            •  Edge computing device + algorithms for monitoring the safety of
                                               miners and mining equipment
                                            •  Hardware, algorithm, platform and service layers

                       Technology keywords  Edge Computing, Surveillance and Monitoring for Mining, AI Stream-
                                            ing Accelerator

                       Data availability    Both publicly and privately available data

                       Metadata (type of    Video streaming input
                       data)

                       Model Training and   Anomaly detection, Diffusion for generation of new samples, CNN,
                       fine tuning          detection, classification and segmentation.

                       Test beds or pilot   •  Baijigou coal mine
                       deployments          •  100,000 alarms with 95% accuracy



                      29�2� Use case description


                      29�2�1  Description

                      According to statistics from Global Energy Monitor (GEM)[1], there are nearly 3 million miners
                      worldwide.  However, the working environment in mining areas is exceptionally complex with
                      frequent occurrences of major production accidents. With the development of the Internet
                      of Things (IoTs), most mining areas have deployed cameras, sensors, and other devices to
                      monitor the safety of production.  Due to limitations such as manpower, distance, and operating
                      environments, these solutions are unable to timely detect and intervene during unsafe behaviors






                  116
   127   128   129   130   131   132   133   134   135   136   137