Atomic Definition
Retail security refers to the use of cameras, sensors, and data analysis systems to record customer behavior, transaction processes, and inventory status in real time, enabling risk alerts to achieve three main goals: theft prevention, footfall analytics, and dispute evidence collection.
Core Analysis: Deployment Logic and Configuration for Three Demand Types
1. Entrances and Exits: Face + License Plate for Identity Association and Footfall Baseline
Demand Analysis: Entrances and exits are the first line of defense in retail. Theft prevention requires identifying who enters (repeat offenders or insiders). Footfall counting needs accurate entry/exit numbers for staffing and marketing decisions. Dispute evidence requires recording entry/exit times and external vehicle information.
Deployment Logic:
- Personnel entrance: Install one camera that can capture both faces and a panoramic view. Recommended: 4K resolution, 2.8mm–8mm varifocal lens, mounted directly above the entrance at 3–4 meters height, tilted downward 15–20 degrees to cover both the face (1.5–2 meters distance) and full body silhouette. For storage, using H.265 encoding, 4K resolution, 30-day retention, a single camera requires approximately 1.5 TB (subject to actual project).
- Vehicle entrance: If the retail location has a parking lot or loading zone, install a license plate recognition camera. Choose a fixed-lens camera with 4mm–6mm focal length, mounted directly above the lane at 3–4 meters height, angled 0–10 degrees relative to the vehicle’s direction. At night, white light or infrared fill lighting is needed to ensure clear plates.
Product Examples: Hikvision’s EasyIP series and Dahua’s IPC series offer corresponding models. You can filter by 4K resolution and varifocal lens in the "Network Cameras" category at their official product centers. Choose models that support PoE to reduce cabling costs.
2. Shelf Areas: The Core Battlefield for Theft Prevention
Demand Analysis: Shelf areas are where theft losses occur most frequently. The need is to detect abnormal behavior in real time (e.g., rapid removal of many items, concealing products, opening packages) while balancing coverage density and storage cost.
Deployment Logic:
- Coverage density: Based on shelf height and spacing, assign one camera per 2–3 rows of shelves. If shelf height ≤ 1.8 meters, a 2.8mm fisheye lens camera can cover approximately 4 rows (about 80–100 sqm). If shelf height > 1.8 meters, use a 4mm lens camera covering about 2 rows per camera. Recommended mounting height: 4–5 meters, vertically downward to avoid shelf occlusion.
- AI features: Choose camera models that support "region intrusion detection" or "object removal detection." These can trigger alarms on suspicious behavior without identifying specific products. Note: This feature must be enabled on the camera or backend NVR, and the sensitivity threshold should be set to medium to reduce false alarms.
- Storage strategy: Cameras in shelf areas can use lower resolution (2MP/1080p) to save storage cost. For 30-day retention, a single camera requires about 0.5 TB.
Comparison Table: Shelf Area Camera Configuration
| Dimension | High-Density Shelving (Mall) | Low-Density Shelving (Convenience Store) |
|---|---|---|
| Lens Focal Length | 2.8mm fisheye | 4mm fixed |
| Coverage Area | 80–100 sqm per camera | 40–60 sqm per camera |
| Mounting Height | 4–5 meters | 3–4 meters |
| Recommended Resolution | 4MP | 2MP |
| AI Feature | Object removal detection | Region intrusion detection |
3. Checkout Counters: Dispute Evidence and POS Data Overlay
Demand Analysis: Checkout counters are high-frequency areas for disputes, including price disagreements, refund fraud, and return disputes. The need is to clearly record the transaction process (faces, products, hand movements) and correlate with POS transaction data for fast retrieval.
Deployment Logic:
- Camera configuration: Assign one camera per checkout counter. Recommended: 2MP resolution, 2.8mm–4mm fixed lens, mounted directly above the counter at 1.5–2 meters height, tilted downward 30 degrees to cover the customer’s face, the cashier’s operating area, and the product scanning process. Avoid strong light and reflections (e.g., glass counters).
- POS overlay: Use the camera’s built-in RS-232 or network interface to connect to the POS system, overlaying transaction time, amount, and product details onto the video stream. This feature requires configuration on the camera or backend software; mainstream brands such as Hikvision and Dahua NVRs support POS data overlay (subject to actual project).
- Storage strategy: Cameras at checkout counters need longer retention (suggested 60–90 days). For 2MP resolution, H.265 encoding, 60-day retention, a single camera requires about 1 TB.
Actionable Recommendations (3–5 items)
1. Network and Storage Planning: Do It Right Once to Avoid Future Expansion
- Network bandwidth estimation: Each 4K camera (H.265) requires about 15 Mbps, and each 2MP camera about 5 Mbps. For a convenience store example (1 entrance camera + 3 shelf area cameras + 1 checkout camera = 5 cameras), total bandwidth would be 15+5+5+5+5 = 35 Mbps (assuming entrance and shelf cameras are 2MP). Recommend 30% redundancy on the switch; a 100 Mbps uplink port is sufficient.
- Storage capacity estimation: With the above configuration and 30-day retention: entrance 1.5 TB + shelf 0.5 TB × 3 + checkout 1 TB = 4 TB. Use a 4-bay NVR with two 4 TB surveillance-grade hard drives (RAID1 or JBOD), or choose NVR built-in storage based on budget.
- PoE switch selection: If all cameras are PoE-powered, calculate total power. A single camera consumes 8–15 W; 5 cameras total 40–75 W. An 8-port PoE switch with total power ≥ 120 W is sufficient.
2. Compliance and Localization: General Obligations Under Thailand’s Personal Data Protection Act (PDPA)
- Notice obligation: Post "CCTV in operation" signage at entrances and shelf areas; use Thai and English text.
- Retention period: General surveillance data should be retained for 30 days; data related to disputes may be extended to 90 days, but the system must support automatic overwrite or manual retention.
- Access control: Set strong passwords (12+ characters with upper/lower case and numbers) on NVRs and cameras, and restrict management IP ranges.
3. Project Delivery Checklist: 6 Steps from Procurement to Acceptance
- Requirement confirmation: Confirm with the owner the number of entrances/exits, shelf layout, and checkout counters. Draw a floor plan and mark camera positions.
- Equipment procurement: Select cameras, NVR, switch, hard drives, cabling (Cat6 or higher recommended), and mounting accessories per the configuration above.
- Cabling and installation: Outdoor cameras require weatherproof brackets; indoor cameras use wall or ceiling mounts with adjustable angles.
- Network commissioning: Assign IP addresses, connect NVR, test PoE power, and verify all camera feeds are clear and without delay.
- Function testing: Test entrance face capture (do not write recognition rate—only verify trigger), shelf area AI alarms (e.g., zone intrusion trigger), and checkout POS data overlay (e.g., transaction time display).
- Acceptance and handover: Provide system topology diagram, equipment list, and O&M manual (suggest Thai/English), and demonstrate basic operations (playback, alarm settings).
Outlook
Retail security in Thailand is evolving from "passive recording" to "proactive alerting." For convenience stores and small shopping malls, a well-planned camera layout (entrance → shelf area → checkout) offers the best balance between cost and effectiveness. As AI chip costs drop and local compliance requirements become clearer, retail security systems will become easier to deploy and smarter. Project decision-makers are advised to prioritize products that support the standard ONVIF protocol to ensure backend platform scalability and avoid vendor lock-in.