1. Introduction: What Should You Consider When Choosing an AI People Counter?
AI people counting technology has become an important part of digital operations in retail stores, shopping malls, transportation facilities, and commercial buildings. By combining 3D vision, artificial intelligence, and data analytics, modern people counters can do more than record how many people enter or leave a space. Depending on the model and configuration, they can also provide occupancy monitoring, visitor demographics, staff exclusion, repeat-visitor analysis, and other operational insights.
However, choosing the right AI people counter is not simply a matter of selecting the device with the highest advertised accuracy.
Different projects have different priorities. A retail chain may need to distinguish customer traffic from employee movements. A shopping mall may require multi-entrance visitor analysis. A transportation facility may prioritize high mounting heights and broad coverage. A system integrator may focus on open interfaces, centralized management, and long-term maintenance.
A suitable selection process should therefore evaluate six key factors:
- Counting performance: Accuracy, bidirectional counting, and performance under real operating conditions.
- AI analytics: Availability of staff identification, gender and age analysis, group counting, and other relevant functions.
- Effective customer traffic analysis: Whether the system can apply business-specific rules to distinguish different types of traffic and reduce duplicate counting.
- Installation requirements: Mounting height, field of view, coverage, environmental conditions, and power supply.
- Data integration: API availability, supported communication protocols, and compatibility with existing business systems.
- Software and total cost of ownership: Analytics platforms, software licenses, deployment costs, maintenance, and scalability.
This guide compares four people counting solutions—FOORIR HX-CCD21, V-Count Nano AI, FootfallCam Pro2, and Xovis PC Series—to help buyers understand their capabilities and select a solution based on actual project requirements.
The comparison is intended as a technical selection guide, not a universal performance ranking. Specifications can vary by model generation, firmware, software license, and installation conditions. Always confirm the exact configuration before purchasing.
2. Overview of Four AI People Counting Solutions
2.1 FOORIR HX-CCD21: AI 3D People Counting for Effective Customer Traffic Analysis
FOORIR HX-CCD21 is an AI-powered 3D people counter designed to provide directional traffic statistics and additional analytics for retail and commercial environments.
The device combines dual-lens stereo vision with AI processing to measure people entering, exiting, passing through, and returning to monitored areas. Depending on the selected configuration, it can also provide gender and age analysis, employee identification, Re-ID-based duplicate reduction, dwell-time analysis, and multi-device deduplication.
These capabilities can help retailers better understand the difference between total people movement and the customer traffic relevant to their business objectives.
For example, employees may enter and leave a store repeatedly throughout the day. Delivery personnel may also contribute to entrance counts without representing shopping visits. If these movements are included in the same metric as customer visits, the resulting data may be less suitable for sales conversion analysis.
HX-CCD21 supports employee filtering and AI-based visitor analysis to help businesses establish more meaningful traffic measurement rules.
Key reference specifications include:
- Counting accuracy: Up to 99%, according to the manufacturer.
- Recommended AI installation height: Approximately 2.5–4.0 m, within the broader supported height range.
- Installation height range: 1.9–6.0 m.
- Horizontal and vertical field of view: 140° and 120°.
- Coverage width: Approximately 1.1–8.4 m, depending on installation height.
- Power: PoE or DC power.
- Connectivity: Ethernet and 2.4 GHz Wi-Fi.
- Data interfaces: HTTP(S), MQTT, FTP/SFTP, and other supported interfaces.
- Environmental protection: IP66.
- Operating temperature: -10°C to 50°C.
- Offline data cache: Up to 90 days.
The manufacturer’s reference figures for gender recognition, age analysis, employee identification, and deduplication are approximately 95%, 80%, 80%, and 80%, respectively. These are separate metrics and should not be confused with the headline people counting accuracy.
Official references: FOORIR HX-CCD21 Product Page and HX-CCD21 Technical Specification.
Best suited for evaluation: Retail projects requiring employee filtering, repeat-visitor analysis, delivery-personnel filtering where supported by the selected configuration, and integration with existing data platforms.
2.2 V-Count Nano AI: AI-Based People Counting for Retail Analytics
V-Count Nano AI is an AI people counting sensor designed primarily for indoor retail entrances and commercial environments.
The manufacturer describes the device as supporting people counting, employee exclusion through an optional solution, gender and age recognition, group counting, occupancy monitoring, and integration with the BoostBI analytics platform.
Its on-device AI processing is designed to produce anonymous analytics rather than relying on a conventional video-recording workflow.
Published reference specifications and features include:
- People counting accuracy: Up to 99%.
- Gender recognition: Approximately 95%.
- Age recognition: Approximately 80%.
- Employee exclusion accuracy: Approximately 90%, with the optional Staff Exclusion Solution.
- Installation height: Approximately 2.2–7.0 m.
- Power supply: 5 V USB-C, with an optional external PoE splitter.
- Connectivity: Wi-Fi.
- Analytics platform: BoostBI.
- Operating temperature: 0°C to 45°C.
The device is intended for indoor installation. Projects involving weather-exposed entrances should evaluate an appropriate outdoor product rather than applying the indoor Nano AI specifications to an outdoor configuration.
For retail operators, the main selection considerations are the required analytics functions, compatibility with the selected software licenses, mounting conditions, and integration with existing systems.
Best suited for evaluation: Indoor retail entrances, visitor demographics, staff exclusion, group counting, occupancy monitoring, and retail analytics through BoostBI.
2.3 FootfallCam Pro2: People Counting and Business Analytics
FootfallCam Pro2 combines 3D vision and AI-based processing for entrance measurement and broader commercial applications.
An important consideration is that the Pro2 name covers different product generations and published specifications. The original Pro2 (2.0), Pro2 2025, and the currently presented Pro2 product configuration should not automatically be treated as identical.
FootfallCam’s published comparison between Pro2 (2.0) and Pro2 2025 lists the following differences:
- Pro2 (2.0): Manufacturer-stated counting accuracy of 95%.
- Pro2 2025: Manufacturer-stated counting accuracy of 99.5%, using sensor fusion involving 3D stereo vision, AI Re-ID, and AI Depth View.
- Pro2 2025 installation height: Approximately 2.1–6.0 m.
- Pro2 2025 coverage area: Approximately 84.79 m² at a mounting height of 4.5 m, according to the comparison page.
- Pro2 2025 processing: Octa-core CPU and 8 TOPS edge AI.
- Advanced analytics: Re-ID, demographic analytics, unique-visitor metrics, and additional in-store analytics, subject to the specified configuration.
The current Pro2 product page also advertises counting accuracy of up to 99.8%. Because the manufacturer’s current product page and historical version comparison present different figures, buyers should confirm the exact hardware generation and applicable accuracy statement before using these numbers in a project specification.
The Pro2 platform can also be evaluated for staff exclusion, occupancy-related applications where supported, data export, and integration with the FootfallCam V9 platform or other supported systems.
Best suited for evaluation: Retail and commercial projects requiring entrance measurement, visitor analytics, Re-ID capabilities in the appropriate version, and integration with a commercial analytics platform.
2.4 Xovis PC Series: 3D People Counting for Diverse Spatial Environments
Xovis PC Series uses 3D vision technology for people counting and movement analysis across retail, transportation, and other public or commercial environments.
The PC Series includes multiple models designed for different mounting heights and installation conditions. Selecting the correct model is therefore essential when planning coverage and counting performance.
The Xovis selection guide provides different installation ranges for different product families. For example, the PC2 family is designed for lower mounting heights, while standard PC3 models are intended for higher installations. Certain specialized models can support mounting heights of up to approximately 20 m.
For projects with large entrances, high ceilings, or complex spatial layouts, the selection process should consider:
- The specific sensor model and its supported mounting height.
- Coverage at the proposed installation height.
- People flow patterns, occlusion, and overlapping counting zones.
- Network connectivity and power requirements.
- Available analytics, software features, and integration options.
It is important not to generalize the installation height, coverage area, environmental protection rating, or accuracy of one Xovis model to the entire PC Series.
Buyers should obtain the relevant product datasheet and installation plan for the exact model being considered.
Best suited for evaluation: High-ceiling installations, transportation facilities, and projects where sensor coverage must be planned around specific spatial and mounting requirements.
3. AI People Counter Comparison: Key Specifications and Capabilities
The following table summarizes selected published specifications and features. It is intended for initial screening rather than a controlled laboratory comparison.
| Comparison item | FOORIR HX-CCD21 | V-Count Nano AI | FootfallCam Pro2 | Xovis PC Series |
|---|---|---|---|---|
| Primary application | Retail and effective customer traffic analysis | Indoor retail analytics | Retail and commercial people counting | People counting across varied spatial environments |
| Core technology | Dual-lens 3D stereo vision and AI | AI-based sensing with on-device processing | 3D vision and AI; configuration-dependent | 3D vision; model-dependent |
| Published counting accuracy | Up to 99% | Up to 99% | Current product page: up to 99.8%; Pro2 2025 comparison: 99.5% | Confirm for the specific model and test conditions |
| Reference installation height | 1.9–6.0 m; AI recommendation approximately 2.5–4.0 m | 2.2–7.0 m | Current product page: 2.1–6.0 m | Varies by model; some PC3 models support approximately 6–14 m |
| Field of view | 140° horizontal, 120° vertical | Ultra-wide; confirm the exact specification | 160° on the current product page | Confirm for the selected model |
| Gender analysis | Approximately 95% reference figure | Approximately 95% | Available in the Pro2 2025 comparison; confirm configuration | Confirm model and software support |
| Age analysis | Approximately 80% reference figure | Approximately 80% | Available in the Pro2 2025 comparison; confirm configuration | Confirm model and software support |
| Staff exclusion | Supported; approximately 80% reference figure for employee identification | Optional Staff Exclusion Solution; approximately 90% reference figure | Available capabilities vary by generation and configuration | Confirm the selected solution |
| Re-ID / repeat-visitor analysis | Supported; approximately 80% reference figure for deduplication | Confirm the required configuration | Supported by Pro2 2025 according to the version comparison | Confirm the selected model and analytics solution |
| Connectivity | Ethernet and Wi-Fi | Wi-Fi | Ethernet | Confirm the selected model |
| Power | PoE or DC | USB-C 5 V; optional PoE splitter | PoE on the published current specifications | Confirm the selected model |
| Environmental rating | IP66 | Indoor Nano AI; separate outdoor model available | Confirm the exact model | Confirm the exact model |
| Offline data cache | Up to 90 days | Confirm retention and operating mode | Confirm the applicable configuration | Confirm the applicable configuration |
| Analytics platform | FOORIR cloud platform and supported interfaces | BoostBI | V9 platform and supported integrations | Xovis analytics solutions |
Important notes on the comparison
- “Up to” figures represent manufacturer-published maximum or reference values, not guaranteed performance at every installation.
- The accuracy figures for different brands may be based on different test methods, sample sizes, traffic conditions, and definitions of a correct count.
- Demographic recognition, employee identification, and Re-ID performance are different metrics from entrance counting accuracy.
- Features may depend on software licenses, optional modules, firmware versions, or additional infrastructure.
- Coverage area must be calculated using the exact model, mounting height, field of view, entrance geometry, and installation plan.
A fair purchasing decision should be based on the specific product configuration and a consistent acceptance test.
4. How to Choose the Right AI People Counter for Your Project
4.1 For Retail Stores That Need More Meaningful Customer Traffic Data
Retailers often use entrance counts to calculate conversion rates, evaluate marketing campaigns, and compare store performance.
However, not every person entering a store is necessarily a customer. Employees, delivery personnel, and visitors who repeatedly cross the counting line can affect the interpretation of total traffic data.
If your objective is to analyze customer visits rather than simply measure all movement, evaluate whether the system supports the relevant filtering rules.
Questions to ask include:
- Can the system identify employees, and what method is used?
- Can delivery personnel be identified under the expected operating conditions?
- Does the system support repeat-visitor analysis or deduplication?
- Can different traffic metrics be reported separately?
- Can the resulting data be integrated with transaction records?
FOORIR HX-CCD21 is worth evaluating when employee filtering, delivery-personnel identification under supported configurations, and Re-ID-based duplicate reduction are important requirements.
V-Count Nano AI and FootfallCam Pro2 can also be considered for staff exclusion and other retail analytics, depending on the selected configuration.
The most appropriate solution is the one that meets the retailer’s required data definitions and performs consistently during a real-world pilot.
4.2 For Visitor Demographics and Retail Customer Insights
If the project requires gender and age analysis, compare the supported demographic functions, recognition performance, reporting tools, and licensing conditions.
FOORIR HX-CCD21 and V-Count Nano AI publish reference figures for gender and age recognition. FootfallCam Pro2 2025 also lists demographic analytics in its version comparison.
For any candidate device, verify:
- Which demographic categories are available.
- Whether the feature is included or requires a separate license.
- The conditions under which the recognition figures were measured.
- Whether results can be segmented by store, time, or traffic direction.
- How the system handles incomplete or uncertain classifications.
Demographic analysis should be treated as an additional source of operational insight rather than a substitute for sales, transaction, or customer feedback data.
4.3 For High Ceilings and Large Coverage Areas
Large shopping centers, transportation hubs, and public buildings often have high ceilings, wide entrances, and complex people flow patterns.
In these environments, the installation height and coverage area can be as important as the sensor’s headline accuracy.
Xovis PC Series includes models designed for different mounting heights, including high-ceiling applications. The current FootfallCam Pro2 specifications also describe coverage for higher installations within its supported range.
FOORIR HX-CCD21 and V-Count Nano AI can be evaluated where their installation ranges and coverage suit the entrance geometry.
Before selecting a device, confirm:
- The minimum and maximum mounting height for the exact model.
- The effective coverage at the proposed height.
- Whether people may be obscured by signs, doors, displays, or other visitors.
- Whether multiple sensors are needed.
- How overlapping zones and duplicate counts will be managed.
- Whether the installation can be maintained safely and efficiently.
A site survey or coverage calculation is preferable to choosing a sensor based on field-of-view figures alone.
4.4 For Challenging Environmental Conditions
Projects involving semi-outdoor areas, exposed entrances, dust, moisture, or temperature variation should evaluate the environmental rating of the exact product version.
FOORIR HX-CCD21 publishes an IP66 rating and an operating temperature range of -10°C to 50°C. V-Count offers a separate outdoor product for weather-exposed applications, while Nano AI is intended for indoor use.
For FootfallCam and Xovis, confirm the environmental specifications of the selected model rather than assuming that all products in a series share the same rating.
Also consider:
- Direct sunlight and changing illumination.
- Water exposure and condensation.
- Ambient temperature and ventilation.
- Stable mounting and cable routing.
- Whether an enclosure or additional protection is required.
An IP rating alone does not determine whether a device is suitable for every outdoor or industrial environment.
4.5 For Integration With POS, ERP, or Business Intelligence Systems
Many organizations want to combine people counting data with transaction records, workforce management, or business intelligence platforms.
In these projects, data integration should be evaluated before purchasing hardware.
Confirm whether the system supports:
- API access and technical documentation.
- Required protocols, such as MQTT or HTTP(S).
- Historical data retrieval and scheduled exports.
- Multi-store aggregation and centralized management.
- Data buffering during network interruptions.
- Access permissions and software licensing.
- Integration with the organization’s existing data architecture.
FOORIR HX-CCD21 lists support for multiple interfaces, including HTTP(S) and MQTT. V-Count provides REST API integration through BoostBI. FootfallCam and Xovis also offer integration options, but exact interfaces and supported functions depend on the selected product and configuration.
A technical integration test can help identify licensing, data-format, and connectivity requirements before a large deployment.
4.6 For Multi-Store Deployment and Long-Term Management
For retail chains and large organizations, the cost of a people counting system includes more than the initial device purchase.
The total cost of ownership may include hardware, installation, network infrastructure, software subscriptions, AI feature licenses, integration, maintenance, and future expansion.
During the evaluation process, ask each supplier to clarify:
- The cost per entrance or monitored zone.
- Which analytics features are included.
- Whether cloud services require recurring fees.
- How remote configuration and firmware updates work.
- How devices are monitored and maintained.
- Whether adding new locations requires additional infrastructure.
- What support is available during deployment and operation.
A clear comparison of these items helps organizations evaluate the long-term suitability of each solution.
5. What Is Effective Customer Traffic, and Why Does It Matter?
Traditional people counting focuses on measuring how many people enter or leave a space. Effective customer traffic analysis goes a step further by applying business-specific rules to determine which traffic metrics are relevant to a particular decision.
Consider a retail store where the entrance counter records all movements. Employees may repeatedly enter and leave during their shifts, while delivery personnel may visit several times a day. Repeat visitors may also cross the same counting line more than once.
These movements can all be valid observations, but they do not necessarily represent unique customer visits.
It is therefore useful to distinguish three concepts.
Total people traffic
The number of people counted crossing a defined line during a specified period.
Filtered traffic
Traffic data after applying configured rules, such as employee exclusion or supported delivery-personnel classification.
Business-defined effective customer traffic
A metric defined by the retailer to support a specific purpose, such as conversion analysis or marketing measurement. Its meaning depends on the rules used and the available data.
These metrics should not be treated as interchangeable.
For example, a repeat visit may be excluded when calculating unique visitors but retained when analyzing visit frequency. Similarly, a delivery worker may be excluded from customer conversion analysis but remain relevant to building occupancy management.
How People Counting Data Supports Conversion Analysis
A common retail metric is the visitor conversion rate:
Conversion Rate = Number of Transactions ÷ Eligible Customer Visits × 100%
The key is to define the denominator consistently.
If the numerator represents completed transactions but the denominator includes employees and non-customer movements, the resulting percentage may not accurately reflect customer conversion.
AI people counters can help retailers refine the data used in these calculations by providing the relevant classification and filtering capabilities. The resulting metric should still be validated against transaction data and the retailer’s own business rules.
For projects that prioritize effective customer traffic analysis, FOORIR HX-CCD21 provides a combination of people counting, employee identification, and Re-ID-related functions that can be evaluated against these requirements.
The best approach is to establish the required metric first, then test whether the selected hardware and software can produce it reliably.
6. How to Validate People Counting Accuracy Before Purchasing
Published accuracy figures are useful for shortlisting products, but a controlled pilot is necessary to evaluate performance in the intended environment.
A practical validation process should include the following steps.
Step 1: Confirm the Exact Product Configuration
Record the model, hardware generation, firmware, software version, and optional licenses.
This is particularly important when a product name covers multiple generations or configurations.
Step 2: Define the Counting Rules
Specify what constitutes an entry, an exit, a repeat visit, and a valid customer visit.
For example, decide how to handle people who turn around at the entrance, employees crossing the counting line, and visitors who pass through multiple monitored zones.
Step 3: Conduct a Reference Count
Use a clearly defined manual counting procedure or another agreed reference method.
Test different periods, including normal traffic, peak hours, bidirectional movement, groups walking together, and situations where people may partially obscure one another.
Step 4: Calculate Accuracy Consistently
A simple absolute error measure is:
Absolute Percentage Error = |Device Count − Reference Count| ÷ Reference Count × 100%
Counting accuracy can then be defined as 100% minus the absolute percentage error for the specified test period, where appropriate.
However, this formula should not be treated as a universal certification method. Projects should agree on the test procedure, minimum sample size, counting direction, and acceptance threshold before testing begins.
For employee identification, demographic analysis, and Re-ID, separate evaluation methods are needed because these functions measure different outcomes.
Step 5: Validate the Data Under Real Operating Conditions
Run the device during representative operating periods and investigate discrepancies.
If the project involves multiple entrances or overlapping monitoring zones, test cross-device deduplication separately.
Document the results so that each supplier is evaluated using the same acceptance criteria.
7. Frequently Asked Questions About AI People Counters
Q1. What is an AI people counter?
An AI people counter is a device that uses sensing technology and algorithms to detect and count people moving through a defined area. Depending on the model, it may also provide directional counts, occupancy data, demographics, staff exclusion, or visitor analytics.
Q2. Is a 3D people counter more accurate than every other counting technology?
Not necessarily in every environment. 3D people counters use depth or stereo information to help distinguish people and understand movement, but performance depends on the device, installation, traffic conditions, and counting rules. The most reliable comparison comes from testing candidate devices under consistent conditions.
Q3. What is the difference between staff exclusion and Re-ID?
Staff exclusion aims to identify employee movements and separate them from other traffic according to the configured rules.
Re-ID, or re-identification, attempts to associate repeated observations of the same visitor within a defined system scope. It can support repeat-visitor analysis and duplicate reduction.
These functions solve different problems and may need to be evaluated separately.
Q4. Can AI people counters identify delivery personnel?
Some systems support classification of delivery personnel under specific configurations, such as uniform or equipment recognition. The capability should be confirmed for the exact model and operating conditions.
If delivery-personnel filtering is a critical requirement, include representative delivery workers in the pilot test.
Q5. Which AI people counter is best for retail stores?
The best choice depends on the store’s objectives.
- FOORIR HX-CCD21 can be evaluated for employee filtering, repeat-visitor analysis, and effective customer traffic measurement.
- V-Count Nano AI can be evaluated for indoor retail counting, staff exclusion, group counting, and demographic analytics.
- FootfallCam Pro2 can be evaluated for entrance measurement and commercial analytics, with careful attention to product generation.
- Xovis PC Series can be evaluated for retail or public-space projects where the specific sensor model fits the installation geometry.
These are application-based starting points, not a universal ranking.
Q6. Can an AI people counter work outdoors?
Some people counters are designed for outdoor use, while others are intended for indoor environments. Check the exact model’s ingress protection rating, operating temperature, mounting requirements, and installation instructions before deploying it in an exposed location.
Q7. Can people counting data be integrated with POS or ERP systems?
Many systems provide APIs or other data interfaces for integration. The available methods, supported data fields, authentication requirements, and licensing conditions vary by supplier and configuration.
Confirm the interface requirements with the supplier and test the integration before deploying the system across multiple stores.
Q8. Does Re-ID necessarily mean facial recognition?
No. Re-ID and facial recognition are not synonymous. The specific data used, processing method, retention policy, and identification scope depend on the implementation.
Organizations should review the supplier’s technical documentation and privacy information to understand how the system processes and stores data.
Q9. How much does an AI people counting system cost?
The total cost depends on the number of entrances, hardware configuration, installation requirements, software licenses, cloud services, and integration work.
For an accurate quotation, provide the supplier with entrance dimensions, ceiling heights, required analytics, expected deployment scale, and integration requirements.
8. Conclusion: Choose an AI People Counter Based on Your Business Requirements
Selecting an AI people counter requires a balance between counting performance, analytics capability, installation conditions, integration requirements, and long-term cost.
FOORIR HX-CCD21, V-Count Nano AI, FootfallCam Pro2, and Xovis PC Series offer different combinations of sensing technology and analytics capabilities. Each should be evaluated against the actual requirements of the intended project.
For retailers who want to move beyond basic entrance counts and establish more meaningful customer traffic metrics, FOORIR HX-CCD21 is worth evaluating for its combination of 3D people counting, employee identification, Re-ID-related analysis, and supported data interfaces.
The recommended selection process is straightforward:
- Define the business metric you need to measure.
- Identify the necessary hardware and AI functions.
- Confirm the installation and integration requirements.
- Compare the exact model specifications and total cost.
- Validate counting performance through a representative pilot.
The goal is not simply to select a device with the highest published number. It is to choose a people counting solution that produces consistent, useful data under the conditions in which your organization operates.
For businesses evaluating AI people counting solutions for retail, commercial buildings, or other people-flow environments, FOORIR can help assess installation requirements, analytics needs, and integration options.
Explore the FOORIR HX-CCD21 AI People Counter to learn more about its technical specifications and effective customer traffic analysis capabilities.
