B-REAL: A keypoint-less behavior platform for offline classification and real-time detection

Abstract

Two obstacles limit automated behavior analysis in neuroethology: the burden of first training pose estimation models, and the inability to identify behavior quickly enough for real-time classification that could trigger stimuli, drug delivery, or environmental manipulations. Here we present B-REAL, a pose-free classifier that maps raw video to behavior faster than video frames are recorded, even on a CPU. Given substantially less training data, B-REAL matched or surpassed established methods trained on full benchmark rodent or fly datasets, particularly on rare and even single-frame behaviors. Pretraining on unlabeled video further improves its accuracy without slowing inference or training time. In closed loop, rewarding mice for B-REAL-detected grooming more than doubled their performance of grooming bouts. B-REAL thus makes closed-loop behavioral experiments practical with minimal annotation and standard hardware.

LTK-YOLO: Lightweight Keypoint Detection for Robotic Tea Bud Picking
Xiaomei Yi, Zekai Zheng, Hao Liang +3
Plants (Basel, Switzerland)1 Oct 2026Peer-reviewedCC BYFull textPlant biologyEurope PMC
A streamlined proof of concept CRISPR-based environmental biosurveillance platform for the detection of closely related invasive rats
B. Duran-Vinet, A. J. Piaggio, N. Foster +8
bioRxiv24 Sep 2026PreprintAll rights reservedFull textEcologybioRxiv
ABEL: an active-learning behavior estimation and labeling platform
J. L. Ritchie, B. E. George, A. V. Roland +12
bioRxiv30 Aug 2026PreprintCC BYFull textNeurosciencebioRxiv
E. coli-produced nanobody fusions as a pan-serotype platform for AAV purification, quantification, and detection
L. Haverkamp, M. Fialho-Gerst, K. M. Müller
bioRxiv28 Oct 2026PreprintCC BY-NC-NDFull textSystems & synthetic biologybioRxiv
Comparing Temporal Averaging and Transformer Encoders for Video-Based Cattle Behavior Classification Under Severe Class Imbalance.
Daniyar Pirlepesov Maratovich, Dae-Yeong Kang, Mohammad Ismail +6
Veterinary sciences10 Oct 2026Peer-reviewedCC BYFull textPlant biologyEurope PMC

Cite

BibTeX RIS

BibTeX

@misc{eagle2026breal,
  title = {{B-REAL: A keypoint-less behavior platform for offline classification and real-time detection}},
  author = {A. Eagle and B. Kohler and V. Kamrowski and N. Lingam and R. Li and E. A. Yttri},
  howpublished = {bioRxiv},
  year = {2026},
  month = oct,
  doi = {10.64898/2026.10.13.758637},
  url = {https://www.biorxiv.org/content/10.64898/2026.10.13.758637v1},
  note = {CC BY},
}

RIS

TY  - UNPB
TI  - B-REAL: A keypoint-less behavior platform for offline classification and real-time detection
AU  - A. Eagle
AU  - B. Kohler
AU  - V. Kamrowski
AU  - N. Lingam
AU  - R. Li
AU  - E. A. Yttri
PY  - 2026
DA  - 2026/10/13
PB  - bioRxiv
DO  - 10.64898/2026.10.13.758637
UR  - https://www.biorxiv.org/content/10.64898/2026.10.13.758637v1
AB  - Two obstacles limit automated behavior analysis in neuroethology: the burden of first training pose estimation models, and the inability to identify behavior quickly enough for real-time classification that could trigger stimuli, drug delivery, or environmental manipulations. Here we present B-REAL, a pose-free classifier that maps raw video to behavior faster than video frames are recorded, even on a CPU. Given substantially less training data, B-REAL matched or surpassed established methods trained on full benchmark rodent or fly datasets, particularly on rare and even single-frame behaviors. Pretraining on unlabeled video further improves its accuracy without slowing inference or training time. In closed loop, rewarding mice for B-REAL-detected grooming more than doubled their performance of grooming bouts. B-REAL thus makes closed-loop behavioral experiments practical with minimal annotation and standard hardware.
ER  -

Abstract

Related

Cite