Highlights of IVS3800

The IVS3800 uses the container-based cloud platform and adopts the high-performance, distributed computing and storage framework to provide users with high-density, dynamic, and elastic shared resource pools. The solution can quickly integrate algorithm management services of third-party algorithms, supports alarming within seconds based on a library, and supports target data clustering. Additionally, the solution supports cross-platform multi-level video management networking and intelligent cross-domain collaboration, and provides unified management and O&M, implementing one cloud across the network.

All-Cloud Synergy

• Data/Service convergence

  • Data convergence: unified storage of video and images

  • Service convergence: convergence of networking management, streaming media forwarding, storage, analysis, and search services

  • Lightweight cloud OS: one server for all services, resource pooling, and on-demand capacity expansion

• On-demand combinations of storage, compute, and search resources, one server for all services, and optimal TCO

  • Applicable to multiple service scenarios of the customer, container technology used, and on-demand combinations of storage, compute, and search resources

  • On-demand deployment of multiple algorithms, one product model with N capabilities, and resource sharing

  • Highly integrated, saving equipment room footprint and reducing overall power consumption

• Deployment of storage, compute, and search resources in cluster mode; elastic scaling

  • Distributed horizontal expansion, linear capability expansion, and unified cluster

  • Dynamic task and data allocation for load balancing

  • Automatic detection of faulty nodes in a cluster and service migration

• Network-wide one cloud, three-level collaboration, supporting cross-domain video collaboration

  • Task collaboration: three-level cross-domain alert and search, supporting unified alert on the entire network; unified cross-domain trajectory search; alert and search based on anonymized features to ensure information security

  • Algorithm collaboration: unified algorithm version management and data search; online algorithm deployment, loading, and upgrade on demand; algorithm update in an upper-level domain, and algorithm download and update in lower-level domains

  • Resource collaboration: three-level cross-domain resource collaboration for centralized analysis, improving resource utilization

    The upper-level domain can borrow resources from lower-level domains. Video files are sliced in the upper-level domain. Distributed parallel analysis is performed in lower-level domains.

    Lower-level domains can borrow resources from the upper-level domain. Live video and recording files of cameras in lower-level domains can be forwarded to the upper-level domain for centralized analysis.
    

  • Data collaboration: on-demand network-wide data aggregation and converged storage
    

    Subscription to data by camera or data dimension (feature, structured data, and full image/partial image) for centralized storage in the upper-level domain, implementing cross-area and cross-system data sharing and supporting data convergence and multi-trajectory integration

Hard Core Innovation

• Extraction and precise recognition of personal and vehicle features

  • Target alerts, person search by image, person tagging, and practical person-related application (SDK interface)

  • Vehicle alerts, vehicle search by image, fuzzy search of license plates, and vehicle application (SDK interface)

• Cloud-based slicing, supporting up to 50x analysis

• GPU for computing acceleration

  • Data match in a target library with millions of records in seconds

  • 1:N target match acceleration: uses the cluster mode to improve the match performance linearly by three times compared with the CPU-based match

  • Long and short features: reduce feature dimensions. Quick match of short features for one-time filtering and secondary precision match of long features improve data match efficiency.

Data Intelligence

• Decoupling of algorithms from applications, resources, and data, integrating high-quality algorithms, and providing unified services

  • Decoupling of algorithms from hardware: refined resource management, resource sharing by multiple algorithms, and on-demand scheduling. Resources are reused despite of algorithm updates.

  • Decoupling of algorithms from applications: unified distributed task management, refined algorithm management, and standard APIs. Algorithm updates do not affect ISV applications.

  • Decoupling of algorithms from data: unified video, image, and structured data. Algorithm change requires only update of the algorithm and features.

  • Decoupling of algorithm plug-ins from the platform version: supports independent release and upgrade of algorithm plug-ins.

• Algorithms provided as apps, accelerating integration of long-tail algorithms in the algorithm repository

  • Unified scheduling and on-demand provisioning of compute resources

  • Unified management of basic information about multiple algorithms

  • Attribute enhancement after algorithm updates; transparent data transmission

  • Unified access and storage of video and images

• Traffic-based scheduling and automatic service resource awareness, improving resource utilization

  • Urgent task: quickly preempts resources of the secondary priority in case of major events

  • Scheduled task: starts real-time or historical data analysis for specified cameras at a specified time

  • Resource reuse during off-peak hours: automatically schedules resources for tasks with low priorities during off-peak hours on the basis of defragmentation

  • Resumable analysis: After a historical image or video analysis task is resumed, the system continues to analyze the task from the breakpoint, ensuring that the task is not repeatedly executed.

• Distributed N:N target clustering, match of millions of records within minutes, promoting practical innovation

Secure and Reliable

• E2E security design specifications, focusing on data security and providing permission control, access control, security control, and operation audit

• E2E reliability design, providing continuous high service availability

• SafeVideo+ and N+0 cluster, ensuring 99.999% reliability of recording services

Easy to Use

• Server transportation with disks, featuring easy deployment

• Cloud-based platform with preinstalled software and hardware, implementing quick service rollout

• Adding and configuring cameras following a wizard

Hope the answer can help you.