Model reference — Live feed
This service counts people crossing a line on a live camera and reports occupancy continuously, so that current numbers are known as they change — for safety limits, door control, staffing and reporting.
Live counting differs from counting recorded footage in one respect that governs everything: the work must keep up. Every frame arrives whether or not the previous one has been processed, so the models must fit inside the frame interval, and if they cannot, frames are dropped and the count drifts. That constraint sets the model sizes, and it is why the practical figure to plan around is how many camera streams one server can carry rather than how fast a single model runs.
Models in this group take a live camera or stream as input and must keep pace with it in real time. Each specification table gives three hardware tiers — Minimum, the smallest setup on which the model runs correctly; Medium, the usual production configuration; and High, a configuration sized for peak volume. The rate is how many camera streams or feeds one server of that tier can keep up with in real time, not a per-hour count: live work must fit inside the interval between frames, and a server that cannot keep pace drops frames rather than falling behind. Use these rates for initial sizing. Before production, benchmark your own data to validate accuracy, latency, throughput and cost. The sample input here is one 1080p stream at 25 frames per second.
Vendor: Ultralytics
What it does: detects people on every live frame, in a range of sizes so the model can be matched to the hardware available. The usual live detector.
| Requirement | Minimum | Medium | High |
|---|---|---|---|
| GPU type | RTX 3060 | RTX 4090 | A100 80 GB |
| VRAM | 6 GB | 24 GB | 80 GB |
| vCPUs | 4 | 8 | 24 |
| RAM | 16 GB | 32 GB | 64 GB |
| Server | 1× RTX 3060 12 GB | 1× RTX 4090 24 GB | 1× A100 SXM 80 GB |
| Rate (camera streams handled at 25 frames per second) | ≈ 10 streams | ≈ 36 streams | ≈ 100 streams |
Vendor: Baidu (PaddlePaddle)
What it does: produces steadier detections without a duplicate-removal step, which gives more predictable timing — valuable when the frame budget is tight.
| Requirement | Minimum | Medium | High |
|---|---|---|---|
| GPU type | RTX 3060 | RTX 4090 | A100 80 GB |
| VRAM | 8 GB | 24 GB | 80 GB |
| vCPUs | 6 | 12 | 24 |
| RAM | 16 GB | 32 GB | 64 GB |
| Server | 1× RTX 3060 12 GB | 1× RTX 4090 24 GB | 1× A100 SXM 80 GB |
| Rate (camera streams handled at 25 frames per second) | ≈ 5 streams | ≈ 18 streams | ≈ 50 streams |
Vendor: Huazhong University of Science and Technology
What it does: follows each person so a crossing is counted once. Light enough to add to a live pipeline at almost no cost, and essential to a correct count.
| Requirement | Minimum | Medium | High |
|---|---|---|---|
| GPU type | RTX 3060 | RTX 4090 | A100 80 GB |
| VRAM | 6 GB | 24 GB | 80 GB |
| vCPUs | 4 | 8 | 24 |
| RAM | 16 GB | 32 GB | 64 GB |
| Server | 1× RTX 3060 12 GB | 1× RTX 4090 24 GB | 1× A100 SXM 80 GB |
| Rate (camera streams handled at 25 frames per second) | ≈ 12 streams | ≈ 40 streams | ≈ 110 streams |
Vendor: Tel Aviv University
What it does: holds a person’s track through brief occlusion, which prevents double counting in busy doorways at the cost of more computing per stream.
| Requirement | Minimum | Medium | High |
|---|---|---|---|
| GPU type | RTX 3060 | RTX 4090 | A100 80 GB |
| VRAM | 6 GB | 24 GB | 80 GB |
| vCPUs | 4 | 8 | 24 |
| RAM | 16 GB | 32 GB | 64 GB |
| Server | 1× RTX 3060 12 GB | 1× RTX 4090 24 GB | 1× A100 SXM 80 GB |
| Rate (camera streams handled at 25 frames per second) | ≈ 6 streams | ≈ 22 streams | ≈ 60 streams |
Vendor: Shenzhen Institute of Advanced Technology
What it does: a very small detector that runs on the camera-side device, allowing counting where no images may be sent to a server at all.
| Requirement | Minimum | Medium | High |
|---|---|---|---|
| GPU type | RTX 3060 | RTX 4090 | A100 80 GB |
| VRAM | 4 GB | 24 GB | 80 GB |
| vCPUs | 4 | 8 | 24 |
| RAM | 8 GB | 32 GB | 64 GB |
| Server | 1× RTX 3060 12 GB | 1× RTX 4090 24 GB | 1× A100 SXM 80 GB |
| Rate (camera streams handled at 25 frames per second) | ≈ 20 streams | ≈ 70 streams | ≈ 200 streams |
Vendor: University of Illinois
What it does: estimates density as a cross-check at peak, when individuals are too tightly packed for tracking to be reliable.
| Requirement | Minimum | Medium | High |
|---|---|---|---|
| GPU type | RTX 3060 | RTX 4090 | A100 80 GB |
| VRAM | 6 GB | 24 GB | 80 GB |
| vCPUs | 4 | 8 | 24 |
| RAM | 16 GB | 32 GB | 64 GB |
| Server | 1× RTX 3060 12 GB | 1× RTX 4090 24 GB | 1× A100 SXM 80 GB |
| Rate (camera streams handled at 25 frames per second) | ≈ 8 streams | ≈ 30 streams | ≈ 90 streams |
The right configuration depends on your camera angles, footfall and how many streams must run on one machine. Our consultants review your views, validate counts against manual observation, and recommend the models, the placement and the hardware for your stream count.
At the start of a project we may run a short proof of concept on a sample of your own data, measuring the accuracy and the throughput the model actually achieves on your material. That replaces the estimates on this page with real figures, so the cost and the schedule for the full engagement are known before it is committed.
Share a representative sample, expected volume, latency target and deployment location for People entrance / exit counter. We benchmark the shortlisted models, recommend the lowest-cost GPU configuration that meets the target, and scale it from pilot capacity to a dedicated production cluster.