Missed Samples vs. Dropped Frames in EthoVision – Causes and Remedies
Last updated: Oct 29, 2026
Problem
Users are confused about the difference between 'missed samples %' and 'subject not found' in EthoVision tracking results, and want to understand why video frames are dropped or unavailable during acquisition.
Details
Affects all EthoVision XT versions. Symptoms include unexpectedly high missed-sample percentages in results tables, gaps in trajectory tracks, or 'subject not found' entries in raw data. Often observed when using high-resolution cameras, high frame rates, or computationally intensive detection methods (e.g., deep learning) on underpowered hardware.
Solution
1. Understand the distinction:
- 'Subject not found' means EthoVision acquired the frame successfully but the detection algorithm could not locate the animal (e.g., animal is hidden, under a shelter, or contrast is insufficient).
- 'Missed samples %' (unavailable samples) means frames were never acquired or processed—EthoVision skipped them entirely.
2. Common causes of missed/unavailable samples:
- CPU/GPU is overloaded by the combination of tracking method, video codec, acquisition, and any simultaneous trial-control tasks.
- Camera resolution or frame rate exceeds what the computer can process in real time.
- Camera is not delivering frames consistently (cable, NIC, or PoE issue).
- Hard drive write speed is insufficient for the selected video quality.
3. Remedies:
- Reduce camera resolution or frame rate to match the computer's processing capacity.
- Switch to a less computationally demanding detection method if deep learning or multiple-body-point tracking is not required.
- Ensure the GigE camera NIC has Jumbo Frames enabled and the dedicated NIC is used exclusively for the camera.
- Use a fast SSD for video storage rather than a spinning hard drive.
- Close all non-essential background applications during acquisition.
- Review EthoVision hardware requirements and upgrade RAM or GPU if the system is below specification.