CCTV & NVR Upgrade Proposal

196-Room Hotel · 25 Storeys · Multi-Level Car Park · Penang, Malaysia

At a Glance — Why This Upgrade Is Needed

The existing system was installed 10 years ago. It cannot identify faces, read plates, or reliably retain footage beyond 30 days.

Current Cameras
105
85% are sub-2 megapixels
Proposed Cameras
141
All 4–8 MP with AI analytics
Face Capture Today
0
No cameras capable
Face Capture Proposed
19
Every lift lobby + entrances
ANPR / Plate Reading
0
Cannot read any plate
ANPR Proposed
11
Entry, exit & every ramp
Current Storage
50 TB
79% full — critically low
Proposed Storage
500 TB
90+ days retention with headroom

Critical Limitations of the Current System

Cannot Identify Anyone

85% of cameras record at 1.2 megapixels and 5 frames per second — too low to identify faces, read plates, or capture detail on fast-moving subjects.

No Number Plate Recognition

Car park entry and exit cameras are 1.2 MP OEM units. It is physically impossible to read a Malaysian number plate from their footage at any distance.

Guest Floor Blind Spots

Every guest floor has only 2 cameras (1 lift lobby + 1 corridor). One entire end of the corridor — leading to the Junior Suite — is unwatched.

Storage Running Out

Synology B is at 81% capacity. Effective retention is around 30 days before footage is overwritten. Both NVR units are end-of-life hardware.

No AI or Smart Detection

Every pixel change triggers recording — rain, shadows, insects. No way to filter for actual people or vehicles. Searching footage is entirely manual.

OEM Cameras — No Support

69 cameras are unbranded OEM units with no manufacturer support, no firmware updates, and unknown remaining lifespan.

What the Upgraded System Delivers

Facial Recognition

Search by face across the entire building. Upload a photo — the system shows everywhere that person appeared with timestamps. VIP alerts and banned-person watchlists.

Number Plate Recognition

11 dedicated ANPR cameras track every vehicle entering, exiting, and moving between car park levels. Search by plate number instantly.

Full Corridor Coverage

Every guest floor gets 4 cameras — face capture at the lift, plus 3 corridor domes covering every metre. Zero blind spots.

90+ Day Retention

500 TB usable storage. Comfortable 90-day retention at full 8 MP quality across 141 cameras with room to grow.

AI-Powered Detection

Every camera classifies motion as person, vehicle, or irrelevant. False alarms from rain and shadows eliminated. Instant event search.

Single-Vendor Ecosystem

100% Hikvision — only 7 camera models across the entire property. Standardised management, easy spares, 10-year platform lifespan.

Sample: ANPR Plate Capture

PFN 8234 ANPR Camera CAPTURED: PFN 8234 15 Jun 2026 — 10:02:34 AM

Dedicated ANPR cameras use infrared strobes and narrow lenses to read Malaysian plates (e.g. PFN 8234) at up to 25 fps — even at night or in dark car park ramps.

Sample: Facial Recognition — Hotel Lobby

MATCH: VIP Guest Scanning... ALERT: Watchlist DeepinView 8MP Face Capture ● 4 faces detected 1 match, 1 alert

DeepinView cameras at lobby entrances capture faces in real time. The NVR matches against VIP and watchlist databases, triggering instant alerts for security staff.

Current System — Why It's Not Enough

Installed 10 years ago. 105 cameras across 2 Synology RS10613xs+ NVRs (both end-of-life).

Fleet Resolution
1.2 MP
85% of cameras — below recognition grade
Frame Rate (OEM)
5 fps
69 cameras at slideshow speed
Storage Used
79%
39.9 TB of 50.6 TB total
Effective Retention
~30 days
Storage runs out before 50-day target

Image Quality — What Security Staff Actually See

Current: 1.2 MP at 5 FPS
1280 x 960 pixels Cannot identify face 02-01-1970 23:14:17
Grainy, blurry, wrong timestamp. Faces are coloured blobs. Useless for identification or evidence.
Proposed: 8 MP at 25 FPS
FACE DETECTED Clear face + AI detection 15-06-2026 17:42:30 3840 x 2160
Sharp detail, correct timestamp, AI face detection. Facial features clearly identifiable. Court-admissible evidence.
Resolution difference: 1.2 MP = 1,228,800 pixels. 8 MP = 8,294,400 pixels. That's 6.75x more detail per frame. The difference between "someone was there" and "this specific person was there at this exact time."

Current Storage — Why Retention Falls Short

Even at the current low bitrates (1.2 MP / 5 fps), the 50.6 TB combined storage struggles to meet the 50-day retention target. With the proposed 8 MP cameras, the current storage would last less than a week.

ZoneCamerasResolutionFPSIssue
Guest Floor Lobbies131.2 MP5No face capture capability at all
Guest Floor Corridors121.2 MP51 camera per floor — blind spot at room 00 end
Car Park151.2 MP5Cannot read any number plate
Lift Cabins51.2 MP5Grainy, poor face detail
Lobby & Entrance10Mixed5–15No dedicated face capture angle
Staircases141.2 MP5Functional but low quality
F&B / Gym / Function26Mixed5–15Mixed OEM and Hikvision hardware
Basement / Pool / Other10Mixed5–15Basic coverage only
Total105

Malaysian Legal Context — CCTV Retention

Malaysia's Personal Data Protection Act (PDPA 2010, amended 2024) governs CCTV footage as personal data. While the PDPA does not mandate a specific number of retention days, it requires organisations to retain footage "only as long as necessary" and delete it securely after. There is no specific Penang state (MBPP) or federal requirement setting a fixed retention period for hotels.

However, the industry standard for hotels across Malaysia and the Asia-Pacific region is 30–90 days minimum. Insurance providers typically require at least 30 days. Law enforcement investigations often request footage up to 90 days after an incident. For a 196-room hotel, 90-day retention is the recommended target to cover guest disputes, insurance claims, and police requests with comfortable margin.

The current system's effective retention of ~30 days meets the bare minimum — but any storage failure or camera addition would push it below that threshold.

Car Park — ANPR Strategy

11 dedicated ANPR cameras for full plate tracking + 12 overview cameras for general surveillance. One ANPR per direction of travel at each ramp.

ANPR Cameras
11
One per direction at each transition
Overview Cameras
12
4MP ColorVu — full colour 24/7
Total Car Park
23
Replaces current 15 (cars + motorcycles)

Multi-Level Car Park — Camera Placement

ANPR cameras are directional — each one faces oncoming traffic. For a two-way ramp, you need one camera per direction. For separate up/down spirals, one camera per ramp section.

ANPR 1Entry — Front-facing

Captures car front plates entering.

ANPR 2Exit — Front-facing

Captures car front plates leaving. Calculates duration.

ANPR 3Entry — Rear-facing

Motorcycle plates. Malaysian motorcycles only have rear plates — this camera faces away from traffic to catch them entering.

ANPR 4Exit — Rear-facing

Motorcycle plates. Catches rear plates of motorcycles leaving the car park.

ANPR 11Front Reserve Parking

Dedicated ANPR for vehicles using the hotel front parking area. Captures plates of delivery trucks, guest drop-offs, and reserved parking users.

L6TOP
Level 6 — Top Floor
Overview: Front Overview: Centre Overview: Back
ANPR 5 Ramp L5→L6 (vehicles going UP to Level 6)    ANPR 6 Ramp L6→L5 (vehicles going DOWN from Level 6)
L5
Level 5
Overview: Front Overview: Centre Overview: Back
ANPR 7 Ramp L4→L5 (UP)    ANPR 8 Ramp L5→L4 (DOWN)
L4
Level 4
Overview: Front Overview: Centre Overview: Back
ANPR 9 Ramp L3→L4 (UP)    ANPR 10 Ramp L4→L3 (DOWN)
L3ENTRY
Level 3 — Entry & Exit Level
Overview: Front Overview: Centre Overview: Back ANPR 1: Entry ANPR 2: Exit

How Vehicle Tracking Works

With 8 ANPR cameras, the system logs every plate at every transition point. Example search result for plate PFN 8234:

TimeCameraEvent
10:02:34ANPR 1 — Entry RampEntered car park
10:03:12ANPR 9 — L3→L4 RampMoving UP to Level 4
10:03:48ANPR 7 — L4→L5 RampMoving UP to Level 5
... (parked on Level 5) ...
14:30:15ANPR 8 — L5→L4 RampMoving DOWN from Level 5
14:30:52ANPR 10 — L4→L3 RampMoving DOWN to Level 3
14:31:20ANPR 2 — Exit RampExited car park — Duration: 4h 28m

Guest Floor Coverage — 4 Cameras Per Floor

14 guest floors (Levels 8–12, 15–23). Each floor gets an identical 4-camera layout based on the actual floor plan dimensions.

Guest Floors
14
Levels 8–12, 15–23 (no 13 or 14)
Cameras Per Floor
4
1 face capture + 3 corridor domes
Total Guest Floor
56
Up from 28 (doubled + no blind spots)
Camera SKUs
2
Only 2 models for all 14 floors

Floor Plan — Camera Positions

Lift lobby is offset west. West corridor is ~11m to room 00 (Junior Suite). East corridor is ~20m to room 17. Current setup leaves the entire west end unmonitored.

Room 00 (JSU) LIFT LOBBY Room 17 (SPD) ◄── 11m west ──► ◄── centre ──► ◄──── 20m east ────► ┌──────────────────┬────────────────────────┬──────────────────────────────┐ │ │ ┌────┐ │ │ │ ◄── Cam C ── │ Cam B ──► │ ◄── Cam A (Face Capture) │ │ 8MP corridor │ │LIFT│ │ │ │ mode (9:16) │ └────┘ │ ◄────── Cam D ──────────► │ │ covers 11m │ │ 8MP corridor mode (9:16) │ │ │ │ looking west from room 17 │ └──────────────────┴────────────────────────┴──────────────────────────────┘ Cam A: 8MP fixed turret 2.8mm — face capture at lift doors, 25 fps Cam B: 8MP varifocal dome 2.8–12mm — corridor east, ~20m coverage Cam C: 8MP varifocal dome 2.8–12mm — corridor west, ~11m to room 00 Cam D: 8MP varifocal dome 2.8–12mm — far end looking back toward lift
Current: 2 Cameras
1 lift lobby + 1 corridor camera. Entire west end (room 00 side) is a blind spot. No face capture.
Proposed: 4 Cameras
Dedicated face capture at lift + 3 corridor domes with overlapping coverage. Every door covered. Zero blind spots.

Guest Floor Coverage — Current Blind Spots vs Proposed Full Coverage

CURRENT: 2 Cameras — Blind Spot
NO CAMERA
CAM
Partial coverage →
Room 00 end (~11m) Toward room 17 (~20m)
1.2 MP at 5 fps — cannot identify faces even in covered area
PROPOSED: 4 Cameras — Full Coverage
← Cam D
← Cam B
Cam C →
Cam D covers 11m A Cam B covers 20m Cam C overlap
8 MP at 15-25 fps — recognition-grade across entire corridor + face capture at lift

Corridor Technical Specifications

Measurements from DXF floor plan and building survey. These dimensions determine camera lens selection, mounting height, and corridor mode configuration.

Total Corridor Length
~31 m
West ~11m + East ~20m
Corridor Width
1.66 m
Narrow — corridor mode (9:16) essential
Corridor Ceiling
2.16 m
False ceiling (bulkhead)
Floor-to-Floor
3.20 m
Structural height between slabs
MeasurementValueImpact on Camera Setup
West stretch (lobby → room 00)~11 mCam D: varifocal set wider (~4mm) — shorter distance = higher pixel density per metre
East stretch (lobby → room 17)~20 mCam B: varifocal set tighter (~6-8mm) — longer distance needs more zoom for recognition grade
Corridor width1.66 mAt 1.66m wide, standard 16:9 wastes 60%+ of frame on walls. Corridor mode (9:16) uses the full sensor vertically
Corridor ceiling height2.16 m (false ceiling)Camera mount height. At 2.16m, face capture turret angle is steep — mount 15cm below ceiling for optimal face angle
Floor slab thickness~150–200 mm~0.84–1.04m above false ceiling to slab above — space for cable runs and dome housings
Room ceiling height~2.7–2.9 mHigher than corridor — the bulkhead drops at corridor junction. Cameras mount in the lower corridor ceiling
Window (east end)Full-height near room 17Cam C faces toward window — consider ColorVu upgrade (DS-2CD2787G2T-LIZS with 140dB WDR) for backlight
Lift lobby positionOffset west of centre3 lifts at centre-west: L1 (guest), L2 (guest), L3 (staff). Service lift L5 at far east end

Storage & Retention Calculator

Interactive calculator with 20% overhead buffer for 24/7 continuous recording, firmware updates, and database indexing.

Total Cameras
141
All recording 24/7 continuously
Aggregate Bitrate
252 Mbps
H.265+ smart codec across all cameras
Daily Recording
2.6 TB
Per day (before overhead buffer)
NVR Capacity
500 TB
2 NVRs × 16 bays × 20 TB drives (RAID 6)

Select Retention Period

Storage includes a 20% overhead buffer for NVR system operations, facial recognition database, ANPR indexing, and filesystem overhead.

Recording Data
234 TB
+ 20% Overhead
47 TB
Total Required
281 TB
Utilisation
56%
281 TB of 500 TB

Comfortable fit with room for future growth.

Storage Comparison — Current System vs Proposed

Camera ProfileCountBitrate EachAggregate30-Day (TB)90-Day (TB)
Face Capture 8MP (25 fps)193.0 Mbps57.0 Mbps18.555.4
Corridor 8MP H.265+ (15 fps)422.0 Mbps84.0 Mbps27.281.6
ANPR 2MP (25 fps)112.0 Mbps16.0 Mbps5.215.6
ColorVu 4MP (15 fps)151.5 Mbps22.5 Mbps7.321.9
Standard 4MP (10–15 fps)491.2 Mbps58.8 Mbps19.157.2
Mini-dome 4MP Lifts (15 fps)51.2 Mbps6.0 Mbps1.95.8
TOTAL (recording only)141255.5 Mbps80.3241.0
+ 20% overhead buffer15.847.5
GRAND TOTAL95 TB285 TB

Network & Switch Upgrade

Standardise all switches to 48-port PoE+ with 10 Gbps SFP+ uplinks. The CCTV network remains physically separate from the hotel data network.

Current Switches
5
Mixed brands, mixed port counts
Proposed Switches
6
All 48-port PoE+ with 10G SFP+
Total PoE Ports
288
vs 141 cameras = 109% spare capacity
Backbone
10 Gbps
Fibre uplinks to server room core
LocationCurrent SwitchCurrent PortsProposed SwitchProposed PortsCameras Fed
Server Room (Core)Cisco SG220-50P
375W PoE, no 10G
48 PoE + 248-Port PoE+ L2+
+ 4× 10G SFP+
48 PoE + 4 SFP+~29
Car Park (L3/L4)None — shares server room48-Port PoE+ L2+
+ 4× 10G SFP+
48 PoE + 4 SFP+20
Level 7TP-Link TL-SG3452XP
Already 48-port + 10G
48 PoE + 4 SFP+48-Port PoE+ L2+
+ 4× 10G SFP+
48 PoE + 4 SFP+~16
Level 12 NewNo switch at this level48-Port PoE+ L2+
+ 4× 10G SFP+
48 PoE + 4 SFP+~24
Level 19 MovedCisco SG220-26P (L15)
26 ports, 180W only
24 PoE + 248-Port PoE+ L2+
+ 4× 10G SFP+
48 PoE + 4 SFP+~32
Level 23 MovedRuijie RG-ES226GC-P (L21/L26)
26 ports each
24 PoE + 248-Port PoE+ L2+
+ 4× 10G SFP+
48 PoE + 4 SFP+~17
TOTAL144 PoE288 PoE141

Port Utilisation — Current vs Proposed

Key Network Improvements

10 Gbps Fibre Backbone

Every floor switch connects to the server room core via fibre SFP+ uplink — no more daisy-chained gigabit copper bottlenecks.

Consistent PoE Budget

Every switch has 370W+ PoE budget. No more worrying about the Level 15 Cisco's 180W limit when adding 8 MP cameras.

Double the Port Capacity

288 PoE ports for 141 cameras = 109% spare capacity. Room to add cameras on every floor without touching the switches.

Standardised Management

All 6 switches are the same brand and model. Single management interface, same firmware, same configuration templates.

NVR Architecture & Monitoring Setup

Two 128-channel DeepinMind NVRs with internal storage, ONVIF support, and unlimited PC client access.

NVR Units
2
DS-96128NI-I16, 128ch each
Internal Drives
32
Seagate SkyHawk AI 20 TB each
Usable Storage
~500 TB
RAID 6, after filesystem overhead
Monitoring PCs
Unlimited
iVMS-4200 — free, any Windows PC

Internal Drive Configuration

SpecDetail
Drive ModelSeagate SkyHawk AI 20 TB (ST20000VE002)
Quantity32 drives total (16 per NVR)
Type3.5" SATA III, 7200 RPM, 256 MB cache
Workload Rating550 TB/year — designed for AI NVR write loads
AI FeatureImagePerfect AI firmware — zero dropped frames during face/plate matching
Vibration SensorsRV sensors for multi-bay environments (critical in 16-bay chassis)
Warranty5 years
Per NVR Raw16 × 20 TB = 320 TB
Per NVR Usable (RAID 6)14 × 20 TB = 280 TB (2 drives parity)
Combined Effective~500 TB after filesystem overhead

Why SkyHawk AI, not regular surveillance drives? Regular drives handle streaming writes but not the random reads that facial recognition and ANPR indexing create. SkyHawk AI has dedicated AI bandwidth that prevents frame drops during simultaneous recording + face matching + plate searching. With 19 face-capture and 11 ANPR cameras running AI, this matters.

System Connection Diagram

SERVER ROOM RACK NVR 1 — DeepinMind DS-96128NI-I16 16× SkyHawk AI 20TB ~70 cameras (guest floors) NVR 2 — DeepinMind DS-96128NI-I16 16× SkyHawk AI 20TB ~71 cameras (car park, lobby, etc) HDMI ×2 Wall Mon. HDMI ×2 Wall Mon. Core Switch (48-Port PoE+ 10G) 10G SFP+ FIBRE via ELV risers SW2 — Car Park 23 cameras ANPR + ColorVu SW3 — Level 7 16 cameras Function + Stairs SW4 — Level 12 24 cameras Floors 8–12 SW5 — Level 19 32 cameras Floors 15–21 SW6 — Level 23 17 cameras Floors 22–25 SW1 — Server Rm 29 cameras B1, Lobby, Lifts MONITORING OFFICES Office A PC + 2 monitors iVMS-4200 Office B PC + monitor iVMS-4200 Hotel LAN to NVRs Remote Access Hik-Connect mobile app HikCentral web browser Access from anywhere ONVIF Compatible Profile S + Profile G Not locked to Hikvision

How Monitoring Works — Not Limited by HDMI

PC

iVMS-4200 (Any PC)

Free client software on every monitoring PC. Multi-monitor support. Live view, playback, face search, plate search. No limit on simultaneous connections.

WEB

HikCentral (Browser)

Web-based VMS — open any browser, no install needed. Floor plan maps, centralised event management, dashboards.

APP

Hik-Connect (Mobile)

iPhone/Android app. Security managers check cameras from anywhere. Works over 4G/WiFi remotely.

TV

HDMI Wall Display

Each NVR has 2× HDMI 4K outputs for passive wall monitors in the server room. Supplementary — not the main interface.

This works exactly like Synology Surveillance Station client today — install on any PC, connect to both NVRs, view all 141 cameras. The upgrade adds face search, plate search, and AI event search that Synology cannot do.

How the 2-NVR Split Works

Unlike your current Synology setup where you log into 2 separate browser sessions, iVMS-4200 connects to both NVRs simultaneously. All 141 cameras appear in one unified camera tree.

NVR 1 — Guest & Upper (71 cameras)
Guest floors L8-L23 (56 cams)
Lift cabins (5)
Lift 5 landings (5)
Level 7 function (5)
90-Day Storage Usage
161 TB of 260 TB (62%)
NVR 2 — Public & Ground (70 cameras)
Car park ANPR + overview (23 cams)
Lobby & entrance (10)
Staircases (18)
F&B, basement, pool, plant (19)
90-Day Storage Usage
122 TB of 260 TB (47%)

Redundancy Logic

If NVR 2 fails, you still have all guest floor recordings (the most sensitive for liability). If NVR 1 fails, you still have lobby face capture, car park ANPR, and entrance coverage. Max retention without expansion: ~150 days.

Your Current Workflow vs Proposed

NOW: 2 Separate Synology Logins
Open browser tab 1 → log into Synology A (56 cameras)
Open browser tab 2 → log into Synology B (49 cameras)
Cannot search across both at once
No face search capability at all
No plate search capability at all
Manual timeline scrubbing only
Finding one person = hours of manual review
PROPOSED: 1 Unified iVMS-4200 Client
Open iVMS-4200 → see ALL 141 cameras from both NVRs
One camera tree, one search, one timeline
Face search: upload photo → results in 2-5 seconds
Plate search: type plate → full entry/exit/floor timeline
AcuSense: filter by person or vehicle events
Install on unlimited PCs in both monitoring offices
Finding one person = under 5 minutes

Future Expansion Options

If retention needs to grow beyond 120 days, each NVR has eSATA ports for external JBOD enclosures (Hikvision DS-A series, 12–24 bays). The NVRs also support iSCSI network storage — so a Synology NAS or any iSCSI target can serve as overflow archive. Additionally, 256 total channels means you can add cameras well beyond the current 141 without touching the NVRs.

Real-World Scenarios — Current vs Proposed

Three actual hotel situations showing how the upgraded system transforms incident response from hours of manual work to minutes of AI-powered search.

Scenario 1: Guest Reports Wallet Stolen from Room

Current System Response
1. Security opens Synology B browser session
2. Finds the corridor camera for that floor
3. Manually scrubs timeline at 5 fps — footage is jerky, faces unreadable
4. Room 00 end has NO camera — if theft occurred there, zero footage
5. Spots a person but can't identify them at 1.2 MP
6. Checks Synology A for lobby camera — separate login, separate search
7. No way to link the corridor person to the lobby footage
Time: 2-4 hours. Result: Usually inconclusive.
Upgraded System Response
1. Security opens iVMS-4200 (one client, all cameras)
2. Pulls up the guest floor corridor — 8 MP, 15 fps, all 4 cameras covering the full corridor including room 00 end
3. Spots a person near the room — face is clearly visible at 8 MP
4. Right-clicks face → "Search by Face"
5. System returns every appearance across ALL 141 cameras in 3 seconds
6. Timeline shows: lobby entry at 14:02 → lift at 14:03 → corridor at 14:05 → left at 14:12
7. Face matched to check-in photo from lobby face-capture camera
Time: Under 5 minutes. Result: Full timeline + identity.

Scenario 2: Vehicle Hits Another Car in Car Park, Drives Off

Current System Response
1. Guest reports damage, provides approximate time
2. Security reviews car park cameras — 1.2 MP, 5 fps
3. Can see a vehicle shape but cannot read the plate at any distance
4. Entry/exit cameras are completely washed out (WDR failure at ramp mouth)
5. No way to determine which floor the vehicle came from
6. Police report filed with "unknown vehicle"
Time: 1-2 hours. Result: Cannot identify vehicle. Guest files insurance claim against hotel.
Upgraded System Response
1. Guest reports damage at Level 5, approximate time 15:30
2. Security reviews Level 5 ColorVu footage — full colour even in dark car park
3. Vehicle colour and shape clearly visible at 4 MP. Impact captured at 15 fps
4. Opens ANPR search → filters by Level 5 ramp between 15:25 and 15:40
5. Gets a list of 4 plates that left Level 5 during that window
6. Cross-references with entry ANPR — identifies vehicle PFN 8234, entered at 12:15, exited at 15:38
7. Exports clip + plate data for police report
Time: 10 minutes. Result: Plate number, entry/exit times, video evidence. Hotel liability eliminated.

Scenario 3: Cash Register Discrepancy — RM 500 Short

Current System Response
1. End-of-day count shows RM 500 shortage
2. Security reviews lobby counter camera
3. Camera shows staff member from the side — can see them standing there
4. Cannot see the register screen, cash drawer, or what their hands are doing
5. Cannot see if cash was given to a guest or pocketed
6. No evidence either way — becomes a he-said-she-said dispute
Time: 1 hour of review. Result: Inconclusive. Cannot prove or disprove theft.
Upgraded System Response
1. End-of-day count shows RM 500 shortage
2. Security opens the overhead register camera — mounted directly above the cash register
3. Can see: register screen, cash drawer, banknotes, and staff hands from directly above
4. Scrubs to each transaction throughout the shift
5. At 16:22 — staff opens drawer, removes RM 500, does not hand it to any guest
6. Side camera confirms no guest at counter at that timestamp
7. Clear evidence for HR action. Exportable clip.
Time: 15 minutes. Result: Clear video evidence from overhead angle. Case closed.

Full Comparison — Current vs Proposed

Every metric improves. The upgrade transforms the system from passive recording to intelligent, searchable surveillance.

MetricCurrent SystemProposed SystemImprovement
Total Cameras105141+36 cameras, zero blind spots
Guest Floor Cameras2 per floor (28 total)4 per floor (56 total)Full corridor + face capture
Dominant Resolution1.2 MP (1280×960)8 MP (3840×2160)6.5× more detail
Frame Rate5 fps (69 cameras)15–25 fpsSmooth, no motion blur
Video CodecH.264H.265+ Smart Codec50–80% smaller files
Facial RecognitionNot possible19 face-capture camerasSearch any face across building
Number Plate RecognitionNot possible11 dedicated ANPR camerasFull plate tracking by floor
AI Motion DetectionNoneAcuSense on all camerasPerson/vehicle classification
Storage Capacity50.6 TB (79% full)500 TB usable10× more storage
Effective Retention~30 days (limited by storage)90+ days comfortably3× longer retention
NVR Platform2× Synology RS10613xs+ (EOL)2× Hikvision DeepinMind 128-chCurrent gen, 10-year lifespan
Camera EcosystemOEM no-name + mixed Hikvision100% Hikvision (7 SKUs)Single vendor, unified management
Network Switches5× mixed (Cisco/TP-Link/Ruijie)6× standardised 48-port PoE+10G backbone, consistent mgmt
PoE Ports Available144 ports (mixed sizes)288 ports (all 48-port)109% spare capacity for growth
Sub-streamNone configuredStream 2 on every cameraFast mobile & live view
Car Park15× OEM 1.2 MP (no plates)12× ColorVu + 11× ANPRFull-colour 24/7 + plate tracking

Camera Capability Comparison