Compare two conditions side by side

Most studies compare conditions: with and without signage, morning and evening, before and after a change to the layout. In CV-Scope each condition is its own experiment. Duplicate the first one and change only the condition, so everything else stays identical.

The practice compares two detectors on the same clip. That is also how you check whether a result depends on the detector.

Step 1. The first condition

1. Open Experiments and open Corridor route choice from guide 11.
2. In Environmental condition notes, type Recommended detector. The notes are stored with the experiment and shown in the comparison.
3. In Tags (comma separated), type detector-test.
4. Click Save.

Step 2. The second condition

1. Click Duplicate at the top of the page. The copy opens, named Corridor route choice (copy).
2. Change Name to Corridor route choice, YOLO11n.
3. Change Environmental condition notes to Detector YOLO11n.
4. In Detector, choose YOLO11n. If it is not in the list, click Save first, install it on the Models page, then come back to the copy and choose it.
5. Click Save.

Only the detector differs now. Camera, scene version, objects to track and rules are the same.

Step 3. Run both

1. Click Start on the copy and wait until its run has completed.
2. Open the first experiment. The comparison adds up all runs of an experiment, so give it exactly one run on the same scene version. If it has a run from guide 11 and you have not changed the scene since, keep that run and do not start another. If you changed the scene, for example with the zones of guide 12, delete the old run in the Runs table, then click Start.

Step 4. Compare

1. Open Analysis.
2. In Compare experiments, tick both experiments.
3. The table shows one row per experiment:

These are the sample results. The route result does not change with the detector. The small detector recorded two more events and a slightly shorter decision time. Small differences like these are normal, and they are why one condition at a time is the rule.

To see all measurements of one experiment, pick it in the Experiment list at the top of the Analysis page.

A real before-and-after study

One live camera, two periods. Run the first experiment during the first period, for example a week without signage. Then duplicate it, give the copy its own Name, change the condition notes, click Save, and run the copy during the second period. Each run belongs to its own experiment, so the periods stay apart.

Two recordings. A scene belongs to a camera, and each recording is its own camera. Copy the scene from one to the other:

1. In the Scene Builder of the first camera, click Export JSON.
2. Open the Scene Builder of the second camera, click Import JSON and choose the exported file. Save.
3. Duplicate the experiment. In the copy, change Name, pick the second camera and click Save.

The two recordings must show the same view. Otherwise the lines and zones do not sit on the same places.

Next

Guide 15 shows where people walked, as a heatmap and as individual paths over the video.