Condition: NOB
RMA Warranty
Description
In order to effectively integrate the video resources of various social organizations and improve the interconnection and interoperability of video resources; share resources and improve the application effect of video intelligent analysis; through integration and integration strategies, open southbound and northbound interfaces, and deeply solve customer application needs, Huawei provides the Huawei Intelligent Video Cloud System with full cloud, full intelligence, three-dimensional network, and wide ecology.
Huawei’s intelligent video cloud platform adopts a full cloud architecture to achieve comprehensive decoupling of software and hardware, data and applications. It uses core technologies such as cloud computing, cloud storage, and big data to build a full-stack cloud-based, full-domain intelligent, and full-cloud collaborative solution that is suitable for industry application scenarios such as safe cities, transportation, and parks.
Product Highlights
Full cloud collaboration
- Computation Pooling
- Use GPU direct access or physical deployment, service on-demand orchestration and management, and accelerate video image analysis.
- Heterogeneous resources are managed in a unified manner to meet different load characteristics.
- Converged Storage
- The new generation of cloud storage architecture, unified storage of multi-source data, and high-speed access to hundreds of billions of data.
- Supports 10TB enterprise hard drives, providing powerful expansion capabilities.
- Redundant coding technology ensures that data is not lost.
- Business Cloudification
- Flexible allocation of resources supports rapid application innovation.
- Businesses are uniformly issued and data centers are uniformly managed for operation and maintenance.
- Cloud-edge collaboration
- Task collaboration: distributed defense and retrieval, unified defense for the entire network; unified search for cross-domain trends; desensitized defense and retrieval of characteristic values to ensure information security.
- Algorithm collaboration: unified management of algorithm versions and unified retrieval of data; online loading and upgrading, the upper-level domain updates the algorithm, and the lower-level domain downloads the updated algorithm.
- Resource collaboration: mutual assistance between superiors and subordinates, centralized resource analysis, and improved resource utilization.
The superior borrows resources from the subordinate, the superior video files are sliced, and the subordinate sites perform distributed parallel analysis;
Subordinates borrow resources from superiors, and subordinate IPC live, recorded, and video files are forwarded to superiors for centralized analysis at the center.
- Data collaboration: Subscribe to the upper-level storage on demand by camera and data dimension (features, structure, size graph), realize cross-region and cross-system data sharing, support data fusion, and integrate multiple tracks.
Hardcore Innovation
- Multiple algorithm warehouses, one platform supports multiple algorithm plug-in integration.
- Algorithms are decoupled from resources, data, and applications, allowing for rapid integration.
- A unified interface is provided to the outside world, and the application layer is unaware of the difference in algorithms.
- Supports online dynamic loading and upgrading of multiple algorithms.
- Flexible combination of multiple algorithms.
- Each algorithm has its own strengths, and more comprehensive value clues can be obtained through parallel retrieval of multiple algorithms.
- Multi-algorithm combined verification, the algorithm performs secondary verification on the defense results, effectively improving the alarm accuracy.
- Cloud slicing, supporting up to 288x speed, and rapid analysis of historical videos.
- Distributed retrieval, billions of feature historical data, and application experience within seconds.
- Distributed retrieval of feature values and balanced storage of hotspot camera data in shards accelerates reading and writing efficiency and avoids traffic congestion.
- Tiered storage, hot data resides in memory, and all data is stored on NVME disks.
- Long and short features, data dimensionality reduction, short features are quickly compared and filtered once, and long features are accurately compared twice to improve comparison efficiency.
- Comparison of distributed portrait libraries, with millions of data points collided within minutes.
- Fine-grained personnel management, high-concurrency retrieval, and step-by-step labeling of the entire population database.
Data Intelligence
- Full view access, the entire network is like a cloud.
- Telecom-class system architecture, supporting up to 500,000 video accesses.
- Adapts to multiple protocols and supports the interconnection of heterogeneous systems.
- Signaling streams and media streams are separated, and the shortest media path is selected to achieve ultra-fast cross-domain video viewing.
- Full-view analysis focuses on feature extraction and accurate identification of people and vehicles.
- Supports target defense, image search for portraits, portrait tagging, and portrait applications (SDK interface).
- Supports vehicle defense, vehicle image search, license plate fuzzy search, and vehicle application (SDK interface).
|
part |
illustrate |
|
|---|---|---|
|
hardware |
2288H V5 |
Used to deploy HUAWEI CLOUD Stack (HCS) and provide computing resources (such as CPU, memory, and network card) for users to create virtual machines on physical servers as upper-layer services (CSP and IVS). |
|
FusionServer G5500 |
GPU hardware physical resources provide GPU acceleration and hard decoding capabilities for business-layer virtual machines. |
|
|
OceanStor 5500 V5/Dorado V3 |
Provide system disk and data disk resources for management and business virtual machines. |
|
|
OceanStor 9000 |
Store video and image data for the business layer. |
|
|
Basic Services |
HUAWEI CLOUD Stack (HCS) |
HCS includes FusionSphere OpenStack and FusionStage.
|
|
CSP |
Provides unified and efficient Web-based visual operation and maintenance functions for upper-layer applications, mainly including application lifecycle management and daily application operation and maintenance. Application lifecycle management includes resource management, application deployment, service governance and grayscale upgrades; daily application maintenance includes application configuration, alarm monitoring, resource monitoring, business monitoring, log management and system management. |
|
|
Business Services |
Image management services |
Provides view access, storage, management and forwarding capabilities. |
|
Image analysis service |
Provide intelligent analysis capabilities. |
|
|
Image retrieval service |
Provides post-analysis retrieval capabilities. |
|
|
Open Services |
SDK Interface |
Provides a northbound open interface for image management services for use by third-party ISVs (independent software vendors). |
|
Scenario-based Restful API |
Provides scenario-based RESTful API interface for image analysis services for third-party ISVs to call. |
|
|
Algorithm interface adaptation package |
Through this algorithm interface, the algorithm package can be independently released and upgraded. |
|
|
API Gateway |
Through the API Geteway open system’s northbound interface, it supports coarse-grained scenario-based combination of open service interfaces. |
|
Mateo (verified owner) –
The product is firmly packed.