Technical Solution Proposal

AIVA Unified AI Video Analytics Platform

This proposal presents AIVA as the core platform product, owned and delivered by Weststar Engineering, with SenseTime and HIK Vision positioned as principal AI video analytics technology partners.

Executive Summary

AIVA as Weststar Engineering’s Product Platform

AIVA is Weststar Engineering’s unified AI video analytics platform layer, built to convert partner AI capabilities into practical monitoring, alerts, search, workflow control, and management reporting.

Owned Product Operational Platform Partner-Enabled

One platform, multiple AI deployment paths

AIVA is designed as the business-facing platform that sits above AI engines and infrastructure. This allows Weststar Engineering to maintain one consistent client-facing product while selecting the most suitable AI partner stack based on project scope, use case, and technical requirements.

Product
AIVA
Unified dashboard, alerting, search, workflows, review, and reporting.
Owned By
Weststar Engineering
Product owner, solution architect, system integrator, and delivery lead.
Platform Relationship View
AIVA Logo
Core Product Platform
Weststar Engineering Logo
Product Owner & Integrator
SenseTime Logo
Principal Partner
HIK Vision Logo
Principal Partner

What AIVA Is

AIVA is Weststar Engineering’s own platform for orchestrating AI video analytics operations through one consistent business-facing interface.

Weststar Engineering’s Role

Weststar owns the product, designs the workflow layer, manages integration strategy, and delivers the implementation model to the client.

Principal AI Partners

SenseTime and HIK Vision extend the AI capability layer, enabling AIVA to support different deployment models while keeping the platform identity unchanged.

Platform Positioning

AIVA Product Stack & Partner Strategy

AIVA should be positioned as the owned business-facing platform, with Weststar Engineering driving solution architecture and delivery, while principal AI partners contribute the underlying analytics capability layer.

Solution Stack View
AIVA Product Layer
AIVA Logo
Unified dashboard, command center, alert workflow, search, reporting, and operational analytics.
↓
Weststar Engineering Layer
Weststar Engineering Logo
Product ownership, solution architecture, implementation logic, workflow design, and delivery.
↓
Principal Partner Layer
SenseTime Logo
HIK Vision Logo
AI recognition, event detection, inference, monitoring services, and deployment-specific AI capabilities.

AIVA Product Layer

  • Unified dashboard and command center
  • Alert workflow and case review
  • Search, reporting, and operational analytics
  • User, role, and access management

Weststar Engineering Layer

  • Owns solution architecture and integration approach
  • Builds business-facing workflows and presentation logic
  • Customizes deployment for each client use case

Principal Partner Layer

  • SenseTime and HIK Vision provide AI video analytic capabilities
  • Capabilities may include recognition, event detection, and AI inference
  • Partner engines are selected or combined based on scope and environment

Recommended Proposal Wording

“AIVA is Weststar Engineering’s proprietary solution platform for AI video analytics operations, enhanced through integration with principal AI technology partners such as SenseTime and HIK Vision.”

Principal Partner Profile
SenseTime Logo

as a Strategic AI Video Analytics Principal

HIK Vision can be presented as a principal AI technology partner that strengthens AIVA with advanced visual intelligence, scalable AI service capabilities, and platform-oriented architecture references such as HIK Vision for operational orchestration concepts.

Partner Value & Background

  • Strong AI and computer vision positioning for large-scale intelligent monitoring scenarios.
  • Suitable as a principal technology partner where advanced analytics and scalable orchestration are needed.
  • Brings architectural thinking that supports complex multi-scene and multi-stream AI operations.

What HIK Vision Represents

  • A platform-style reference for multi-dimensional sensing and operational intelligence.
  • Useful as a model for layered architecture, data intelligence, and orchestration.
  • Supports the idea of separating user interface, analytics engine, middleware, and scene applications.

Why It Matters to AIVA

  • Helps AIVA position itself above the analytics engine as the operational control layer.
  • Supports future expansion into more advanced monitoring, search, alarm handling, and data-driven workflows.
  • Strengthens enterprise proposal credibility where layered architecture and orchestration are important.

Potential Contributions into AIVA

Advanced CV Multi-scene AI Layered Architecture Alarm Orchestration Data Intelligence Search & Review

HIK Vision strengthens the AI capability layer and enterprise-scale architectural direction that AIVA can operationalize.

Recommended Positioning in Proposal

Position HIK Vision as a principal AI technology partner whose platform and analytics strengths enhance AIVA’s ability to deliver scalable, intelligent, and operationally structured AI video analytics solutions.

Principal Partner Profile
SenseTime Logo

as a Strategic AI Video Analytics Principal

SenseTime can be presented as a principal AI technology partner that provides practical, deployment-ready AI video analytics capabilities, operational tooling, and integration-ready platform features that complement AIVA’s business-facing product layer.

Partner Value & Background

  • Strong fit for practical AI video analytics deployment scenarios.
  • Provides operationally relevant AI capability building blocks that can be integrated into project delivery.
  • Suitable for environments requiring configurable monitoring, management, and API-driven extensibility.

What SenseStudio Represents

  • A convergent AI enablement platform for computer vision operations.
  • Brings person, device, space, and monitoring management concepts into one manageable framework.
  • Supports deployment flexibility, administration, policy setup, and integration workflow design.

Why It Matters to AIVA

  • Gives AIVA strong operational AI building blocks for real projects.
  • Supports rapid implementation through practical AI service domains and open interfaces.
  • Improves AIVA’s ability to convert AI outputs into dashboards, workflows, and management reporting.

Operational Core

  • Person management
  • Device management
  • Space / zone management
  • Access control support

AI Capability Layer

  • Recognition workflows
  • Monitoring events
  • Attendance-related support
  • Operational alert scenarios

Deployment Strength

  • Flexible deployment models
  • Standalone / cluster options
  • RTSP camera compatibility
  • Open API approach

AIVA Benefit

  • Faster integration path
  • Stronger implementation confidence
  • Better practical feature coverage
  • Enterprise-ready workflow support
Partner Contribution Strategy

How Principal Partners Strengthen AIVA

AIVA remains the core Weststar Engineering product platform, while principal partners contribute complementary AI strengths that increase technical depth, deployment flexibility, and proposal credibility.

HIK Vision Contribution Pattern

  • Strengthens high-level AI platform and orchestration positioning.
  • Supports enterprise-scale architectural storytelling.
  • Useful for projects emphasizing multi-scene intelligence, layered design, and advanced operational structure.
  • Helps AIVA position itself as a serious top-layer command and orchestration platform.

SenseTime / SenseStudio Contribution Pattern

  • Strengthens practical deployment and integration readiness.
  • Supports real-world management of people, devices, spaces, and monitored events.
  • Useful for projects emphasizing implementation speed, operational workflows, and configurable AI services.
  • Helps AIVA deliver more directly usable project capabilities faster.

AIVA Keeps Ownership

AIVA remains the main product platform owned by Weststar Engineering, including dashboards, workflows, search, review, reporting, and client-facing operational control.

Partners Extend Capability

Principal partners extend the AI capability layer, allowing AIVA to adapt based on project scope, use case, and technical fit without changing the business-facing platform identity.

Commercial Message

This gives Weststar Engineering a stronger story: one owned product platform, supported by credible principal AI partners, with flexibility to match different project demands.

Technical Reference Comparison

HIK Vision vs SenseTime Infrastructure Reference

Category HIK Vision Reference SenseTime Reference AIVA Interpretation
Positioning Project-specific high-capacity hardware requirement provided for deployment planning. Standard product reference sizing from SenseStudio for standalone / cluster video analysis scenarios. HIK Vision appears positioned as a heavier customized node, while SenseTime provides clearer standard sizing baselines.
CPU 2 × Intel Xeon Gold 5416S
2.0GHz, 16 cores / 32 threads each 		2 × Intel SP 4314 for video analysis servers Both are dual-socket server designs, but the HIK Vision requirement is framed as a higher-end enterprise build.
Memory 16 × 16GB RDIMM = 256GB total 256GB DDR4 (standard video analysis server reference) RAM footprint is broadly aligned at 256GB for video analysis workloads.
GPU Strategy 2 × NVIDIA Ampere A16 GPUs
PCIe, 250W, listed as 64GB, passive cooling 		NVIDIA Tesla A2 or T4
1–4 GPUs per server depending on deployment model 		SenseTime explicitly documents A2/T4 as supported reference GPUs. HIK Vision’s requirement is more customized and should be validated against the intended workload design.
Storage Layout 1 × 480GB SATA SSD
23 × 2.4TB SAS 10K hot-swap HDD
PERC H755 SAS RAID controller 		Standalone: 2 × 960GB SSD RAID 1 + 5 × 10K SATA RAID 5
Cluster: 2 × 480GB SSD RAID 1 (system), 2 × 1TB SSD RAID 1 (middleware), 3 × SSD for image data 		HIK Vision’s proposed storage is significantly denser and suggests heavier retention / data volume expectations.
Power Dual hot-swappable 2400W redundant PSUs 1000W or above ×2 (1+1 spare) for SenseTime video analysis servers HIK Vision sizing indicates a materially larger power envelope, consistent with a heavier server profile.
Scaling Reference Not explicitly tied to a published stream-count baseline in the provided requirement. Standalone: up to 100 streams
Standard 3-node cluster: up to 300 streams
High-channel cluster (>3 nodes): up to 800 streams
(subject to algorithm performance spec) 		SenseTime gives clearer published deployment scaling guidance. HIK Vision should be positioned as custom-sized per project capacity.
Operational Note Requires vGPU software for VDI usage Documentation emphasizes video analysis, event handling, and AI inference deployment references HIK Vision’s listed node may be intended to support broader virtualization / presentation needs in addition to analytics workloads.

Recommended Commercial Positioning

Present HIK Vision as a partner that can support larger, more customized enterprise-grade deployment profiles where sizing is driven by project-specific architecture. Present SenseTime as a partner with clearer documented product baselines for standardized video analysis deployment.

Recommended Technical Positioning for AIVA

AIVA should remain hardware-agnostic at the product layer. This allows Weststar Engineering to align the partner stack based on deployment objectives: standardized product-led rollout versus custom high-capacity infrastructure design.

Infrastructure Signal

The supplied HIK Vision sizing indicates a heavier enterprise deployment profile, including dual-socket compute, 256GB memory, dual GPUs, and dense SAS storage. This suggests HIK Vision can be positioned well for customized, higher-capacity infrastructure-led deployments.

Technical Reference Comparison

HIK Vision vs SenseTime Commercial Reference

Commercial Comparison Reference

Commercial Area HIK Vision Reference SenseTime Reference AIVA / Weststar Positioning
Infrastructure Cost Profile Higher initial infrastructure profile based on the supplied heavy server sizing (dual CPU, high-density storage, dual GPU, 2400W PSU). More standardized reference sizing through published SenseStudio deployment models (standalone / cluster / high-channel cluster). HIK Vision may be positioned as a higher-capex custom deployment path, while SenseTime can be positioned as a more standardized baseline-driven deployment path.
Scalability Cost Predictability Likely more project-dependent, with sizing and cost varying based on custom architecture decisions. Easier to estimate using documented stream-based reference deployments. SenseTime may be easier for early budgeting; HIK Vision may require more custom quotation alignment.
Licensing Consideration Additional software cost may apply depending on GPU virtualization / vGPU requirements and deployment model. Licensing is typically aligned to product modules, features, and deployment scale. Keep AIVA’s pricing separate from partner engine licensing so the commercial model stays modular.
Deployment Model Impact More suitable for custom-built enterprise projects where hardware is sized specifically for the client. More suitable for productized deployment planning using standard reference nodes. Weststar can position HIK Vision for tailored enterprise builds, and SenseTime for faster standardized rollout options.
Commercial Story in Proposal “Custom-sized for higher-capacity or more specialized deployment requirements.” “Reference-based product deployment with clearer baseline sizing and scaling guidance.” AIVA remains the main owned platform; partner cost can be aligned based on project scale, required AI scope, and commercial fit.

HIK Vision Cost Tendency

Higher Capex GPU Heavy Storage Heavy

The HIK Vision sizing profile points toward a more customized and infrastructure-intensive commercial path.

SenseTime Cost Tendency

Baseline-Friendly Reference-Led Predictable Scaling

SenseTime’s documented sizing is easier to position as a standardized starting point for budgeting and phased rollout.

AIVA Commercial Positioning

Platform Separate Partner Modular Flexible Proposal

Keep AIVA platform pricing separate from partner infrastructure and AI-engine costs so proposals stay modular and easier to tailor.

Commercial Reference

Market Price Comparison Reference

Component HIK Vision Reference SenseTime Reference Current MYR Cost Estimate Recommended Sell Price (+20%)
CPU 2 × Intel Xeon Gold 5416S 2 × Intel Xeon Silver 4314 Xeon Gold 5416S ≈ RM3,673 each
2 units ≈ RM7,346

Xeon Silver 4314 ≈ RM3,082 each
2 units ≈ RM6,163 		Xeon Gold 5416S ≈ RM4,407.60 each
2 units ≈ RM8,815.20

Xeon Silver 4314 ≈ RM3,698.40 each
2 units ≈ RM7,395.60
Memory 16 × 16GB RDIMM = 256GB 256GB reference sizing 256GB Target by Module Count

16 × 16GB ≈ RM990 each
Total ≈ RM15,840

8 × 32GB ≈ RM3,851 each
Total ≈ RM30,809

4 × 64GB ≈ RM4,909.99 each
Total ≈ RM19,639.96

2 × 128GB ≈ RM18,215 each
Total ≈ RM36,430 		16 × 16GB ≈ RM1,188 each
Total ≈ RM19,008

8 × 32GB ≈ RM4,621.20 each
Total ≈ RM36,970.80

4 × 64GB ≈ RM5,891.99 each
Total ≈ RM23,567.95

2 × 128GB ≈ RM21,858 each
Total ≈ RM43,716
System SSD 1 × 480GB SATA RI SSD 480GB / 960GB SSD-based system references appear in standard deployment sizing Dell MY reference for 480GB SATA RI SSD ≈ RM7,408.78 each ≈ RM8,890.54 each
GPU 2 × NVIDIA Ampere A16 NVIDIA A2 / T4 / L40 / L40S HIK Vision Reference
NVIDIA A16 ≈ RM16,015 each
2 units ≈ RM32,031

SenseTime / Market Reference Options
NVIDIA A2 ≈ RM2,023 each
NVIDIA T4 ≈ RM2,237–RM2,683 each
NVIDIA L40 ≈ RM119,980 each
NVIDIA L40S ≈ RM156,882 each 		HIK Vision Reference
NVIDIA A16 ≈ RM19,218 each
2 units ≈ RM38,437.20

SenseTime / Market Reference Options
NVIDIA A2 ≈ RM2,427.60 each
NVIDIA T4 ≈ RM2,684.40–RM3,219.60 each
NVIDIA L40 ≈ RM143,976 each
NVIDIA L40S ≈ RM188,258.40 each
Bulk Storage 23 × 2.4TB SAS 10K drives More standardized SSD / smaller RAID reference layouts in documented deployment models Major storage capex block
Final pricing depends on selected drive brand, RAID policy, and supplier quote. 		Apply 20% markup on finalized supplier total for the full storage stack.
Power / Platform Envelope Dual 2400W redundant PSU Lower baseline PSU expectations in standard productized sizing Higher PSU rating signals a larger total hardware envelope rather than a standalone pricing decision. Apply 20% markup on finalized PSU / chassis platform quote.
Use Case Layer

Recommended AIVA Solution Use Cases

Attendance & Presence Use Case

Attendance & Presence

Supports face-based presence visibility, attendance workflows, and exception monitoring.

  • Classroom / zone attendance visibility
  • Presence trend dashboards
  • Exception highlighting and daily reporting
Behavior & Incident Monitoring Use Case

Behavior & Incident Monitoring

Helps identify risky or unusual behavior patterns for safety review and operational response.

  • Fighting / aggression review workflows
  • Smoking, climbing, loitering, and intrusion alerts
  • Alarm review with evidence linkage
Search & Investigation Use Case

Search & Investigation

Enables operators to search historical events and perform structured review workflows.

  • Historical event search
  • Image or attribute-based lookup
  • Trajectory / profile style review
Device & Space Operations Use Case

Device & Space Operations

Supports management of cameras, spaces, zones, and operational configuration.

  • Camera inventory and health overview
  • Zone / area configuration
  • Policy assignment by device or location
Alert Command Center Use Case

Alert Command Center

Centralizes alert visibility, escalation tracking, and event management workflows.

  • Real-time alert panel
  • Historical alarm statistics
  • Escalation, status, and ownership tracking
Integration Workflow Use Case

Integration Workflow

Extends AIVA into a connected operational platform through system integration and workflow handoff.

  • External system event push
  • Third-party orchestration hooks
  • Business process handoff after detection
MOE Use Case

Smart School AI Video Analytics Use Cases

Student Relationship Mapping

Attendance & Presence Use Case

Identify recurring student encounters based on repeated detection within the same timestamp and camera context.

  • Tracks repeated student-to-student proximity encounters
  • Flags relationship strength when encounters exceed 50 repeated detections
  • Can be visualized using a relationship chart / network graph

Student Attendance via CCTV

Attendance & Presence Use Case

Automate attendance using CCTV feeds as long as the student is detected within the school compound.

  • Detects student presence from school entry / compound cameras
  • Updates attendance automatically once valid presence is confirmed
  • Reduces dependency on manual roll call

Fence Climb Detection

Attendance & Presence Use Case

Detect behavior traits associated with students attempting to leave the school by climbing fences.

  • Monitors perimeter / restricted boundary zones
  • Flags climbing or escape-like movement patterns
  • Helps prevent student run-off incidents

Fighting Detection

Attendance & Presence Use Case

Detect physical conflict behavior and generate immediate alerts for response teams.

  • Identifies possible fight activity from live video feeds
  • Can alert security guards, discipline teachers, or administrators
  • Supports faster intervention and evidence review

Smoke & Fire Detection

Attendance & Presence Use Case

Detect smoke or fire directly from CCTV streams and notify responsible personnel.

  • Monitors for visible smoke / flame patterns
  • Triggers alerts to guards or relevant authorities
  • Improves emergency response time inside school compounds

Loitering & Group Detection

Attendance & Presence Use Case

Detect lingering student circles and identify possible group congregation behavior.

  • Measures loitering duration in selected zones
  • Counts people within the same area
  • Marks more than 5 people as a group for monitoring

School Safety Value

Improves real-time visibility into safety incidents, perimeter breaches, and risky student behavior.

Operational Value

Reduces manual monitoring effort by turning CCTV feeds into actionable alerts, presence records, and review workflows.

MOE Positioning

AIVA can be positioned as a smart school monitoring platform that supports attendance, student welfare, and campus security under one operational layer.

Solution Architecture

Proposed AIVA High-Level Flow

AIVA is Weststar Engineering’s productized operational and presentation layer, designed to sit above partner AI engines, device inputs, and external systems so the business-facing platform remains stable even when the underlying analytics provider or model strategy changes.

01

Video / Input Layer

RTSP cameras, uploaded images, offline files, NVR feeds, and supported external sources.

→
02

AI / Analysis Layer

Face, human, vehicle, structural analysis, crowd counting, fighting detection, fall-down, fire/smoke, and other configured analytics.

→
03

AIVA Control Layer

Rules, dashboards, event review, search center, user management, approvals, and reporting.

→
04

Business / Integration Layer

External systems, notifications, exports, workflow triggers, and downstream operational actions.

Core Principle

AIVA stays as the stable product layer while partner AI engines can vary by project scope.

System Logic

Detection flows upward into operational control, then outward into reporting and business workflows.

Access
Role-Based
User segmentation for admin, operator, reviewer, and management.
Integration
REST + Push
API-driven integration with webhook-style event forwarding.
Storage
Evidence
Event records, captures, logs, and review metadata.
Deployment
Flexible
On-premise/cloud and standalone/cluster strategies.
Implementation Readiness

Algorithm Boundary & Acceptance Considerations

For proposal accuracy, it is important to present that analytics performance depends on environmental and installation conditions.

Camera / Scene Conditions

Behavior & Incident Monitoring Use Case
  • Recommended light intensity: above 50 lux
  • Camera height guideline: 2.5m–3.5m (common indoor reference)
  • Recommended viewing angle: 15°–30°
  • Suggested horizontal distance: more than 3m
  • Avoid severe backlight, heavy blur, excessive noise, and major occlusion

Face / Registration Quality

Behavior & Incident Monitoring Use Case
  • Eye distance should be at least 60 pixels; more is better
  • Face should be centered, clear, and not heavily occluded
  • Eyes should remain visible; reflective / blocking accessories reduce reliability
  • Longer face presence in frame improves recognition effect

Representative Accuracy Notes

Face Detection ≥ 95% Recognition TP ≥ 90% FP ≤ 1%

These figures are only meaningful when the documented boundary conditions are met.

Proposal Guidance

  • Always recommend a pilot using representative camera positions.
  • Separate platform capability from scene-specific tuning requirements.
  • Document environmental assumptions in the final scope.
Open Integration

API & External System Integration Model

The API references show a broad integration surface that supports account auth, people, devices, policies, events, files, image analysis, and asynchronous processing.

Integration Workflow

  1. Create an external account
  2. Assign allowed APIs to that external account
  3. Request an access token
  4. Invoke platform APIs using tokenized requests
  5. Receive data / events or subscribe via HTTP push

Key API Domains

Person API Device API Monitor Policy Gate Control Event Handler Utility API Cognitive API Async Analysis

Practical AIVA Integration Opportunities

Area How AIVA Can Use It
Person / Group Sync user entities, watchlists, groups, and profile metadata.
Device / Camera Read camera inventory, map devices to zones, and manage monitoring scope.
Events Display live incidents, search historical records, and compute alert statistics.
File / Image Upload Enable manual evidence upload, offline parsing, or review workflows.
HTTP Push Forward system events into custom workflows, notifications, or ticketing logic.
Async Analysis Queue heavier analysis jobs for large files or batch processing scenarios.
Delivery Plan

Implementation Phases & Closing Notes

The recommended execution model is a phased rollout: validate requirements first, prove performance in real conditions, then productize dashboards, workflows, and integrations.

1
Discovery & Positioning

Phase 1

  • Clarify use cases
  • Confirm target users
  • Define success criteria
2
Pilot & Validation

Phase 2

  • Validate camera positions
  • Test boundary conditions
  • Tune detection expectations
3
AIVA Build Layer

Phase 3

  • Dashboard screens
  • Event workflows
  • Search / reports
4
Integration & Handover

Phase 4

  • API integration
  • User acceptance
  • Go-live / support
Key Strength

Reusable Core Platform

AIVA remains Weststar Engineering’s reusable core product platform, while SenseTime and HIK Vision serve as principal AI video analytics partners whose capabilities can be integrated based on project scope, technical fit, and deployment needs.

Key Risk

Expectation Management

Over-promising algorithm performance without documenting camera, lighting, and scene constraints can create avoidable delivery and acceptance issues.

Recommended Next Step

Finalize Proposal Package

Convert this deck into the formal proposal baseline, then refine it with project-specific diagrams, costs, deployment assumptions, and pilot scope details.

AIVA by Weststar Engineering Sdn Bhd