A Framework for Fast, Large-scale, Semi-Automatic Inference of Animal Behavior from Monocular Videos

Price, E.; Khandelwal, P. C.; Rubenstein, D. I.; Ahmad, A.

doi:10.1101/2023.07.31.551177

Home/Archive/bioRxiv

bioRxiv · 10.1101/2023.07.31.551177

A Framework for Fast, Large-scale, Semi-Automatic Inference of Animal Behavior from Monocular Videos

Price, E.·Khandelwal, P. C.·Rubenstein, D. I.·Ahmad, A.

View original source ↗Export metadata ↓

Abstract

An automatic, quick, accurate, and scalable method for animal behavior inference using only videos of animals offers unprecedented opportunities to understand complex biological phenomena and answer challenging ecological questions. The advent of sophisticated machine learning techniques now allows the development and implementation of such a method. However, apart from developing a network model that infers animal behavior from video inputs, the key challenge is to obtain sufficient labeled (annotated) data to successfully train that network - a laborious task that needs to be repeated for every species and/or animal system. Here, we propose solutions for both problems, i) a novel methodology for rapidly generating large amounts of annotated data of animals from videos and ii) using it to reliably train deep neural network models to infer the different behavioral states of every animal in each frame of the video. Our methods workflow is bootstrapped with a relatively small amount of manually-labeled video frames. We develop and implement this novel method by building upon the open-source tool Smarter-LabelMe, leveraging deep convolutional visual detection and tracking in combination with our behavior inference model to quickly produce large amounts of reliable training data. We demonstrate the effectiveness of our method on aerial videos of plains and Grevys Zebras (Equus quagga and Equus grevyi). We fully open-source the code1 of our method as well as provide large amounts of accurately-annotated video datasets2 of zebra behavior using our method. A video abstract of this paper is available here3.

Explore related subjects

animal behavior and cognition Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Price, E., Khandelwal, P. C., Rubenstein, D. I., Ahmad, A.. 2023-08-02. A Framework for Fast, Large-scale, Semi-Automatic Inference of Animal Behavior from Monocular Videos. https://doi.org/10.1101/2023.07.31.551177

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Deconstructing Boldness: Latent behavioural structure across novelty, predation, and exploration in an endangered species translocation program

Risk-related behaviours including neophobia, exploration, and predator avoidance are often assumed to covary along a general boldness axis. Although some translocation studies have linked one or more of these behaviours with fitness, it is possible that the underlying behavioural structure varies under different selective environments. Conservation breeding programs provide a unique opportunity to evaluate the structure of behavioural correlations in a non-domestic species under artificial conditions. We tested whether risk-related behaviours represent a unified boldness syndrome in a captive population of the endangered Attwaters prairie chicken (APC; Tympanuchus cupido attwateri). We quantified the repeatable behaviours of 257 juveniles and compared 11 models of behavioural covariation structure using multivariate mixed models and confirmatory factor analysis. There was no support for a single risk-taking axis; instead, we found three dimensions representing responses to novelty and environmental stimuli (Engagement), predator avoidance (Threat Response), and maze exploration (Maze Progression). Our results indicate that these risk-related behaviours are not organized along a single axis in captive APCs, challenging assumptions that risk-taking represents a universal behavioural syndrome. To anticipate post-release success accurately, conservation managers should understand the exact behavioural architecture of the captive population rather than assume that the classical bold-shy structure occurs in all contexts.

animal behavior and cognition↗

bioRxiv2026

Shared brain transcriptome signatures of social evolution and plasticity in shell-dwelling cichlid fishes

Inter-individual variability in phenotypic traits is widespread in nature both across and within species, with similar patterns of variation often occurring between and within species. Given that these two types of variation are generated by evolutionary and plasticity processes respectively, it raises the question to what extent the causal mechanisms underlying such variability at these two different levels are shared or not. To address this question, we studied the variation in brain transcriptomes in Tanganyikan cichlids that exhibit both intra- and interspecific variation in social behavior. We compared two facultatively social cichlid species, each containing naturally occurring solitary and social-living individuals, with a closely related obligatory solitary resembling the ancestral state. Differential gene expression and Gene Set Enrichment Analyses were performed in the telencephalon and diencephalon transcriptomes to investigate the molecular basis of both behavioral plasticity and the evolution of sociality. Behavioral plasticity involved few differentially expressed genes but extensive changes in biological pathways, supporting a highly polygenic architecture of social behavior. Across both species, social-living individuals consistently upregulated pathways involved in translation, protein homeostasis, and energy metabolism, whereas solitary individuals showed enrichment of neuronal morphogenesis, cell adhesion, and structural plasticity. Remarkably, these same functional programs were associated with the independent evolution of sociality, revealing a strong overlap between evolutionary and plastic transcriptional responses. This convergence was largely restricted to the telencephalon, while the diencephalon showed limited and inconsistent functional signatures. Together, our results support the repeated recruitment of gene regulatory programs during the evolution of sociality and demonstrate that behavioral plasticity and evolutionary transitions share common neurogenomic responses, consistent with a conserved genetic toolkit for social behavior.

animal behavior and cognition↗

bioRxiv2026

Inductive bias anchors a scaling law for biological learning

The speed of animal learning varies substantially across individuals and task domains, and does not always track intuitive notions of task complexity. A nominally simple binary decision task can take thousands of rewards to reach proficiency, whereas seemingly complex naturalistic tasks can be learned within dozens. A long-standing explanation is that learning speed reflects the alignment between an animal's inductive biases and the task at hand. However, without a precise notion of alignment, it is difficult to translate this idea into quantitative predictions. Here, modeling an animal as a Bayesian learner that improves through iterative reward feedback, we use statistical learning theory to derive a scaling law for biological learning. The law makes two predictions. First, inductive bias multiplicatively sets the speed of learning through a single quantity, the effective complexity. This quantity measures how well the task aligns with the animal's prior, and corresponds to the number of bits required to represent the task solution under that prior. We demonstrate this multiplicative scaling in a human motor learning dataset, where behavioral measurements before learning can precisely predict subsequent learning speed. Second, task performance improves as a power law in experience. We validate this power law in a large-scale mouse learning dataset and show that a two-parameter rescaling collapses individual learning curves onto a universal function predicted by the theory. These results suggest that behavioral measurements before learning can be used to predict task-dependent learning speeds, and they explain why learning speed reflects alignment with prior knowledge rather than nominal task complexity.

animal behavior and cognition↗