AUTODOME Gen6 CPP13 vs CPP16 Tracking a Target Through a Crowd

In this evaluation, Steve and the team compared auto tracking performance between two Bosch (KEENFINITY Group) AUTODOME Gen6 variants: the CPP13 model (NDP 7602-Z40) and the newer CPP16 model (NDP 7802-Z40). The goal was simple and very practical: see how each camera handles one of the toughest real world moments for PTZ tracking, when the person being tracked walks into a larger group, everyone mixes together, and then people break back apart.

That scenario is exactly where auto tracking can either look brilliant or fall apart fast.

The test setup in plain terms

They started both cameras tracking the same individual, “Sam.” Sam then approached a small group of people. The group mixed around and shifted positions, then separated again.

While watching the video, the team pointed out two helpful visual cues:

Everyone remained classified as a person.
The tracked subject was identified by a distinct highlight, described as a blue outline, while other people in the scene had a different outline color.

This made it easy to see whether the camera stayed “locked” on Sam or accidentally jumped to someone else once bodies overlapped and crossed.

What happened when Sam entered the group

The key difference showed up quickly.

On the CPP13 camera (the older platform), the tracking became confused as Sam blended into the crowd. The camera changed targets, effectively switching from Sam to Alex. In other words, it did what many auto tracking systems do under pressure: it stayed tracking a person, but not the right person.

On the CPP16 camera, tracking stayed consistent. Even when Sam entered the group and people crossed paths, Sam remained highlighted as the target on the right side view, and the camera maintained tracking on him through the mix up and separation.

The team emphasized the significance here. This is not just a small quality improvement. It is the difference between: “We captured the incident and followed the subject of interest” and “We have video, but the camera tracked the wrong person when it mattered most.”

Smoother tracking and better framing on CPP16

Beyond target retention, they also noticed a motion and framing improvement on CPP16.

They described the CPP16 tracking movement as smoother, with less up and down motion and less jitter. The camera kept the tracked person better composed in the frame, while still preserving more of the surrounding scene.

That point matters more than it sounds. Good auto tracking is not only about staying on the correct target. It is also about producing usable video:

If the camera overcorrects constantly, operators lose confidence in what they are seeing.
If the camera frames too tightly, you lose context like who is near the subject, what direction they are moving, and what else is happening around them.
If it frames smoothly with context, it is easier for a security team to interpret the situation and respond.

In their words, keeping more of the scene in view can “lend itself” to better accuracy because the camera is capturing more context while it tracks.

Why Steve connected this back to IVA capability

Steve tied the tracking results back to the broader theme they have seen across other comparisons: CPP16 shows stronger capability from an IVA perspective, including performance improvements that show up in challenging conditions like distance and complex motion.

Auto tracking in crowds is a great stress test for analytics. It forces the system to handle:

Occlusion, when one person blocks another
Multiple similar objects, such as several people dressed similarly
Rapid target handoff risk, when the system “snaps” to the wrong person
Scene complexity, where the camera must keep identity continuity even when people overlap

In this evaluation, CPP16 handled those challenges more reliably.

Practical takeaway for end users and integrators

If your use case includes active monitoring, guard tours, live response, or investigations where maintaining tracking on a specific person matters, this is exactly the kind of performance difference that can justify choosing the newer platform.

Based on what they observed:

CPP13 can lose the target and switch to the wrong person when the subject blends into a group.
CPP16 stayed on the intended target through the crowd interaction.
CPP16 tracking looked smoother and more stable, producing video that is easier to watch and use, with better scene context.

A quick note on scope

This was a targeted scenario test, not a full lab benchmark. But it is a scenario that mirrors real security environments: lobbies, hallways, campuses, public spaces, and anywhere people naturally cluster.

And in that moment, the CPP16 AUTODOME Gen6 showed a clear advantage in auto tracking behavior and video usability.