Two Mice Merge into One Detection or Shelter/Object Area Causes Detection Loss – Multi-Animal Tracking Optimization
Last updated: Nov 01, 2026
Problem
When two unmarked mice are in the arena together and come into close contact or rest together, EthoVision detects them as a single blob and loses individual identity. Separately, detection near a shelter or in-arena object causes the animal to be lost or misdetected.
Details
Affects experiments measuring social contact, proximity, and time near shelters (e.g., NOR objects, shelters). Observed in EthoVision XT 17 and likely other versions when using standard center-point or contour-based detection with two unmarked animals. The merging problem worsens with poor lighting uniformity or when the background reference contains animals.
Solution
1. Ensure the background reference image is captured with no animals in the arena and under identical lighting conditions to acquisition.
2. Improve lighting uniformity so the full body of each animal has consistent contrast against the floor.
3. Tighten the minimum and maximum subject size settings so that a 'merged blob' of two mice falls outside the maximum size range and is flagged rather than tracked as one animal.
4. Consider color-marking animals (e.g., with non-toxic fur dye) to allow color-based identity discrimination in EthoVision's multi-animal tracking mode.
5. For proximity and contact measures, use the 'Distance between subjects' parameter and define a contact threshold appropriate for mouse body size rather than relying solely on zone entry.
6. Near shelters or objects: ensure the object is excluded from the detection area using the arena drawing tools (doughnut/cutout method) so the animal is not lost when it moves under or beside the object.
7. Review the contour-based detection algorithm as an alternative to center-point detection; it often handles partial occlusion near objects more reliably.