AI cameras for bank surveillance are becoming a mandatory security standard at financial institutions in Vietnam. Combining facial recognition, automatic weapon detection, and 24/7 ATM monitoring, this system not only reduces risks but also optimizes operational costs. The case study below describes the actual deployment process at a mid-sized joint stock commercial bank, with specific measurable data.
Customer Problems

- Lobby security risks: In the first 6 months of 2024, 8 incidents were recorded of individuals wearing masks and full-face coverings entering the lobby. The old camera system (Analog 2MP) could not perform facial recognition, leading to 3-5 days of manual cross-referencing with customer data.
- Manual ATM monitoring: 45 ATM stations spread across locations, each with only a basic recording camera. In 2023, there were 12 ATM break-in attempts and 6 masked cash withdrawal incidents. Average detection time was 8-12 hours after the incident.
- High security personnel costs: The bank hired 15 security monitoring staff (3 shifts/day, 5 people/shift) at a cost of 2.5-3 billion VND/year. However, the real-time incident detection rate was only 35%.
- Lack of system integration: Cameras, fire alarms, and access control operated independently. When an incident occurred, staff had to check 3-4 different screens, taking 15-25 minutes to confirm an alert.
Solution Deployed by An Ninh Số


After 3 months of surveying and design, An Ninh Số deployed an integrated AI camera bank surveillance system in 5 steps. The solution uses AI facial recognition cameras in the lobby, weapon detection cameras at doors, and 24/7 ATM monitoring cameras with infrared sensors.
- Step 1 — Survey and topology design (4 weeks): Measured 25 branches, identified 185 priority installation points: main lobby (48 points), teller counters (62 points), ATMs (45 points), vaults (12 points), employee entrances/exits (18 points). Bandwidth calculation: each 8MP camera requires 12-15 Mbps, total storage capacity 180 TB (90-day retention).
- Step 2 — Install AI cameras in lobby and counters: Used Hikvision DS-2CD2387G2-LSU/SL (8MP AcuSense) at main lobby — facial recognition at 3-8 meters distance, 98.5% accuracy (per internal testing). Dahua IPC-HFW5849T1-ASE-LED (8MP WizMind Full-Color) cameras at teller counters — detection of foreign objects (weapons, unusual backpacks) with under 0.5 second latency.
- Step 3 — Install 24/7 ATM monitoring cameras: Dahua IPC-HFW2439S-SA-LED (4MP Full-Color) cameras at each ATM station, combined with infrared sensors for night mode. Integrated vibration sensors and tamper door sensors — triggering real-time alerts to the VMS.
- Step 4 — Integrate VMS and alarm system: Used Genetec Security Center as the single VMS platform, integrating 185 cameras + 45 ATM sensors + 25 access control readers. Connected with BriefCam RESPOND for real-time analytics: mask detection, unusual crowd gathering (≥4 people in 2 minutes), abandoned objects.
- Step 5 — Training and trial operation (2 weeks): Trained 15 monitoring staff on VMS interface, alert handling procedures (3-level SOP). Ran parallel trial operation with the old system for 14 days, recording 42 test alerts with a 96% correct detection rate.
Achieved Results

- 90% — Reduction in security incidents compared to the same period last year (from 34 incidents down to 3 in 12 months)
- 45% — Reduction in monitoring personnel costs (from 2.8 billion down to 1.5 billion VND/year by reducing 7 staff)
- 30% — Increase in post-incident video search speed (from 4-6 hours down to 1-2 hours thanks to AI search)
- 15 incidents — Detection of individuals wearing masks/unusual accessories in the first 6 months (all received real-time alerts)
"Before, whenever there was an ATM incident at night, it took us at least 8-12 hours to detect it. An Ninh Số's AI camera bank surveillance system completely changed the process: alerts reached my phone within 30 seconds, along with real-time images. After 12 months, we saved nearly 1.5 billion VND in operational costs."
Mr. Nguyen Van Hung, Director of Operations Division, ABC Joint Stock Commercial Bank
Lessons Learned and Application for Other Businesses
Related articles
Case Study: Việt Đức Trí Group deploys 3D AI volumetric measurement cameras with ±2mm accuracy
6 min readAI Safety Surveillance Cameras for Factory Labor Safety: Solution to Reduce Accidents by 70%
7 min readStart with a free 60-minute site survey
Our senior technical team will visit your site, assess the current setup, and propose a bundle matched to your business model and budget.


