Cameras That Not Only See, but Know Where — Seeking IPCAM and NVR Vendors to Co-Create Video × Positioning Convergence

Video knows who is present and what they are doing; positioning knows which room they are in and how far away. P-Square has both AI vision recognition and wireless positioning, supplying camera and NVR vendors with exactly the piece video analytics cannot deliver on its own.

2. Industry Focus: The Ceiling of Video Analytics, and a Maturing UWB Ecosystem

The structural limits of pure video analytics: occlusion, cross-camera identity and privacy

Edge AI cameras are now widespread, but video analytics has three structural limits no better model can remove. A person occluded by machinery, racking or a door simply disappears from frame. Moving to another camera requires re-identification, where error rates remain stubbornly high. And facial recognition is restricted by regulation and privacy policy in many environments.

Questions video cannot answer

"Which room is that wheelchair in?" Video can only report that a wheelchair is in frame, not which room number it occupies.
"Where is this person now?" Tracking breaks the moment they leave the field of view.
"Who entered this area?" Identity is required, but facial recognition may not be permitted.

A maturing UWB ecosystem has lowered the barrier to convergence

UWB has moved from professional into consumer markets: products such as Apple AirTag and Samsung SmartTag have driven module volume and brought costs down. The FiRa Consortium is developing the UWB ecosystem in close cooperation with IEEE, the Wi-Fi Alliance and the Car Connectivity Consortium, enabling new fine-ranging use cases. In parallel, Bluetooth 6.0 introduces Channel Sounding (phase-based ranging), bringing precise ranging to low-cost, low-power BLE.

What this means for camera vendors

Falling module cost: UWB and BLE positioning modules are no longer priced as specialist equipment.
Settled standards: IEEE 802.15.4z and FiRa specifications ensure cross-brand interoperability.
Differentiation headroom: the vast majority of IPCAMs still sell video alone, so positioning is a clear differentiator.

3. Value-Add Pain Points for Camera Vendors, and What P-Square Offers

Taiwanese IPCAM, NVR and IPC vendors generally have complete hardware and manufacturing capability; what is missing is the positioning algorithm and field know-how.

What P-Square provides:

Both video and positioning from a single source: P-Square has AI vision recognition (dual-camera skeletal recognition, dynamic licence plate recognition, people counting, biological recognition) alongside wireless positioning (BLE / Wi-Fi / UWB). Our AI Box already supports multiple camera streams with an optional UWB positioning module — a ready-made reference architecture.
Tags cover the three blind spots of video: Occlusion: radio signals need no line of sight. Cross-camera: a tag ID is inherently unique, so no re-identification is needed. Privacy: tags are represented by a number, so identity mapping needs no biometric data.
Hardware independence and standard integration: The algorithm is not tied to a chipset and supports BLE 4.0/5.0/6.0 Channel Sounding, Wi-Fi RSSI and UWB (ToA / PDOA). A RESTful API delivering JSON connects directly to existing NVR, VMS or operations systems with no back-end rewrite.
A cost-efficient deployment design: The proprietary Locator extension design lets one unit drive four read heads covering five points, cutting hardware volume to about a fifth. Where a site is already installing ceiling cameras, both can be done in one pass.

4. Flexible Partnership Models: Co-Creating a Win-Win

We do not sell modules to vendors. We co-create — you own the hardware, P-Square provides the positioning algorithm, and we go to market together:

Partnership model	Who it suits	What it involves
USB dongle / M.2 module integration	Existing IPCAM / NVR / IPC vendors	No mainboard or antenna redesign — an add-on module gives current models positioning capability immediately, the fastest route to a value-added product line.
Camera onboard integration	New-generation model developers / chipset vendors	Embed the positioning algorithm and software modules into the camera or NVR system for a differentiated model integrating video and positioning in one device.
AI Box joint selling	System integrators and solution providers	Use the P-Square AI Box as the edge node for video analytics and positioning convergence, shipped alongside your existing cameras to form a complete solution quickly.
Joint government project applications	Listed and pre-IPO companies, video majors, system integrators	Combine both parties' strengths to apply for government positioning, AI and smart-facility grants, advancing development while reducing R&D risk.

5. Frequently Asked Questions

We already have AI video analytics — why add wireless positioning?

Because the two answer different questions. Video knows who is present and what they are doing, but breaks the moment someone is occluded, leaves the field of view or moves to another camera. Wireless positioning knows which room they are in and how far away, needs no line of sight, and tag IDs are inherently unique so no re-identification is required. In practice the two are complementary rather than alternatives.

Will adding positioning raise product cost significantly?

It depends on the technology and integration route. A USB dongle or M.2 module requires no mainboard or antenna redesign, so the added cost is limited; where a site only needs room-level accuracy, low-cost BLE is sufficient and UWB is unnecessary. Because our algorithm is hardware independent, the radio technology can be chosen to match the product's positioning.

How does position data reach our existing NVR or VMS?

P-Square provides a RESTful API delivering JSON, so position and event data can be pushed to existing systems in real time; iframe embedding is also supported, so no back-end rewrite is needed.

Does this partnership mean buying modules from P-Square?

No. Our position is that the partner owns the hardware and P-Square provides the algorithm and software, with both parties selling together or applying jointly for government grants. Taiwan's domestic market is limited and international channel costs are high, so sharing development and channel cost is the sensible model.

What has P-Square actually built in image recognition?

Deployed applications include dual-camera AI skeletal recognition (detecting movement correctness from front and side simultaneously), high-precision dynamic licence plate recognition (no stopping required, direction detection, multi-floor support), bus seat and occupancy analysis, and a fish species recognition system. The system supports RTSP IP cameras with six or more streams on a single host.

Discuss a Partnership

Or email service@p2-positioning.com · Tel +886-3-666-9151