Knowledge Base

Problem

When using FaceReader's multi-subject (multi-face) feature to process videos containing multiple people, the software gets stuck at the 'Adding detected participants' step after video processing completes and never generates the participant list.

Details

This is a performance or memory issue rather than a hard capacity limit. It occurs more frequently with longer videos, videos containing many face detections, or scenes where participants move in and out of frame frequently, causing fragmented tracking. Progressively worse behavior on subsequent analyses in the same session may also indicate memory accumulation. Observed in FaceReader 10.

Solution

1. Save your project and fully close FaceReader, then reopen it before attempting further multi-subject analyses — this clears accumulated memory from prior processing.
2. If working with long videos, consider splitting the video into shorter segments and processing each segment separately.
3. Reduce complexity where possible: if many individuals appear briefly, consider whether they need to be tracked, as high face-detection counts increase processing load.
4. Ensure the PC meets FaceReader's recommended hardware specifications, particularly RAM and GPU VRAM requirements.
5. If the issue recurs consistently, note the video duration, approximate number of faces, and available system RAM/VRAM and contact Noldus Support for further investigation.
6. Note: the FaceReader API cannot be used in multi-face analysis mode; API-based data retrieval is only available for single-subject analysis.