Anipose: A toolkit for robust markerless 3D pose estimation

Karashchuk, Pierre; Rupp, Katie L.; Dickinson, Evyn S; Walling-Bell, Sarah; Sanders, Elischa; Azim, Eiman; Brunton, Bingni W.; Tuthill, John C.

doi:10.1016/j.celrep.2021.109730

Cited by 162 publications

(160 citation statements)

References 97 publications

Supporting

Mentioning

132

Contrasting

Order By: Relevance

“…This would be consistent with the higher threshold we found for infants compared to adults. These possibilities point to the need for a longitudinal study across infancy and early childhood, utilizing motion tracking to capture all infant movements ( Karashchuk et al, 2021 , Mathis et al, 2018 , Nakano et al, 2020 , Pereira et al, 2018 ), and EEG alternatives such as newly developed, cryogen-free MEG sensors ( Boto et al, 2018 , Boto et al, 2017 , Boto et al, 2016 , Iivanainen et al, 2019 , Iivanainen et al, 2017 , Knappe et al, 2014 ) which have to potential to allow acquisition of high SNR data from developmental populations.…”

Section: Discussionmentioning

confidence: 99%

Detection and analysis of cortical beta bursts in developmental EEG data

Rayson

Debnath

Alavizadeh

et al. 2022

Developmental Cognitive Neuroscience

View full text Add to dashboard Cite

Developmental EEG research often involves analyzing signals within various frequency bands, based on the assumption that these signals represent oscillatory neural activity. However, growing evidence suggests that certain frequency bands are dominated by transient burst events in single trials rather than sustained oscillations. This is especially true for the beta band, with adult ‘beta burst’ timing a better predictor of motor behavior than slow changes in average beta amplitude. No developmental research thus far has looked at beta bursts, with techniques used to investigate frequency-specific activity structure rarely even applied to such data. Therefore, we aimed to: i) provide a tutorial for developmental EEG researchers on the application of methods for evaluating the rhythmic versus transient nature of frequency-specific activity; and ii) use these techniques to investigate the existence of sensorimotor beta bursts in infants. We found that beta activity in 12-month-olds did occur in bursts, however differences were also revealed in terms of duration, amplitude, and rate during grasping compared to adults. Application of the techniques illustrated here will be critical for clarifying the functional roles of frequency-specific activity across early development, including the role of beta activity in motor processing and its contribution to differing developmental motor trajectories.

show abstract

Section: Discussionmentioning

confidence: 99%

Detection and analysis of cortical beta bursts in developmental EEG data

Rayson

Debnath

Alavizadeh

et al. 2022

Developmental Cognitive Neuroscience

View full text Add to dashboard Cite

show abstract

“…These findings were enabled by a new technical tool–marker-free holistic 3D pose reconstruction and tracking of individual body parts of freely moving animals. Rather than triangulating 2D pose reconstructions (Günel et al, 2019; Mathis and Mathis, 2019; Karashchuk et al, 2021; Nath et al, 2019), FreiPose directly reconstructs body poses in 3D, resulting in higher tracking accuracy than that achieved using previous tools. Analyzing the problem holistically by fusing information from all views into a joint 3D reconstruction allowed us to surpass the commonly used methods by 49.4% regarding the median error in freely moving rats.…”

Section: Discussionmentioning

confidence: 99%

“…Second, thus far, marker-free analyses have mostly been applied in 2D settings (Graving et al, 2019; Mathis et al, 2018; Pereira et al, 2019), which do not pose the need for detailed pose reconstruction of freely moving animals covering all three spatial dimensions. When recording with multiple cameras, 2D outputs can be triangulated to reconstruct a 3D pose a posteriori (Günel et al, 2019; Mathis and Mathis, 2019; Karashchuk et al, 2021; Nath et al, 2019), but such post-processing suffers from the ambiguities in the initial 2D analysis, reducing both accuracy and reliability (Fig. 1 b left).…”

Section: Introductionmentioning

confidence: 99%

3D pose estimation enables virtual head-fixation in freely moving rats

Schneider

Zimmermann²,

Alyahyay

et al. 2022

Preprint

View full text Add to dashboard Cite

The impact of spontaneous movements on neuronal activity has created the need to quantify behavior. We present a versatile framework to directly capture the 3D motion of freely definable body points in a marker-free manner with high precision and reliability. Combining the tracking with neural recordings revealed multiplexing of information in the motor cortex neurons of freely moving rats. By integrating multiple behavioral variables into a model of the neural response, we derived a virtual head-fixation for which the influence of specific body movements was removed. This strategy enabled us to analyze the behavior of interest (e.g., front paw movements). Thus, we unveiled an unexpectedly large fraction of neurons in the motor cortex with tuning to the paw movements, which was previously masked by body posture tuning. Once established, our framework can be efficiently applied to large datasets while minimizing the experimental workload caused by animal training and manual labeling.

show abstract

“…We validated pose trajectories by comparing each pose estimation model’s output to our manual annotations of each participant’s pose (Table 5 ). While manual annotations are susceptible to human error 55 , they are often used to evaluate markerless pose estimation performance when marker-based motion capture is not possible 30 , 56 . We used root-mean-square (RMS) error averaged across all keypoints to evaluate model performance for the 950 frames used to train the model as well as 50 annotated frames that were withheld from training.…”

Section: Technical Validationmentioning

confidence: 99%

AJILE12: Long-term naturalistic human intracranial neural recordings and pose

et al. 2022

Self Cite

View full text Add to dashboard Cite

Understanding the neural basis of human movement in naturalistic scenarios is critical for expanding neuroscience research beyond constrained laboratory paradigms. Here, we describe our Annotated Joints in Long-term Electrocorticography for 12 human participants (AJILE12) dataset, the largest human neurobehavioral dataset that is publicly available; the dataset was recorded opportunistically during passive clinical epilepsy monitoring. AJILE12 includes synchronized intracranial neural recordings and upper body pose trajectories across 55 semi-continuous days of naturalistic movements, along with relevant metadata, including thousands of wrist movement events and annotated behavioral states. Neural recordings are available at 500 Hz from at least 64 electrodes per participant, for a total of 1280 hours. Pose trajectories at 9 upper-body keypoints were estimated from 118 million video frames. To facilitate data exploration and reuse, we have shared AJILE12 on The DANDI Archive in the Neurodata Without Borders (NWB) data standard and developed a browser-based dashboard.

show abstract

Anipose: A toolkit for robust markerless 3D pose estimation

Cited by 162 publications

References 97 publications

Detection and analysis of cortical beta bursts in developmental EEG data

Detection and analysis of cortical beta bursts in developmental EEG data

3D pose estimation enables virtual head-fixation in freely moving rats

AJILE12: Long-term naturalistic human intracranial neural recordings and pose

Contact Info

Product

Resources

About