-
Notifications
You must be signed in to change notification settings - Fork 8
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
5 changed files
with
240 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,240 @@ | ||
<!doctype html> | ||
<html lang="en"> | ||
|
||
<head> | ||
<!-- Required meta tags --> | ||
<meta charset="utf-8"> | ||
<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"> | ||
|
||
<!-- Primary Meta Tags --> | ||
<title>Learnable latent embeddings for joint behavioural and neural analysis</title> | ||
<meta name="title" content="Learnable latent embeddings for joint behavioural and neural analysis"> | ||
<meta name="description" content="Mapping behavioural actions to neural activity is a fundamental goal of neuroscience. As our ability to record large neural and behavioural data increases, there is growing interest in modeling neural dynamics during adaptive behaviors to probe neural representations. In particular, neural latent embeddings can reveal underlying correlates of behavior, yet, we lack non-linear techniques that can explicitly and flexibly leverage joint behavior and neural data. Here, we fill this gap with a novel method, CEBRA, that jointly uses behavioural and neural data in a hypothesis- or discovery-driven manner to produce consistent, high-performance latent spaces. We validate its accuracy and demonstrate our tool's utility for both calcium and electrophysiology datasets, across sensory and motor tasks, and in simple or complex behaviors across species. It allows for single and multi-session datasets to be leveraged for hypothesis testing or can be used label-free. Lastly, we show that CEBRA can be used for the mapping of space, uncovering complex kinematic features, and rapid, high-accuracy decoding of natural movies from visual cortex."> | ||
|
||
<!-- Open Graph / Facebook --> | ||
<meta property="og:type" content="website"> | ||
<meta property="og:url" content="https://cebra.ai/"> | ||
<meta property="og:title" content="Learnable latent embeddings for joint behavioural and neural analysis"> | ||
<meta property="og:description" content="Mapping behavioural actions to neural activity is a fundamental goal of neuroscience. As our ability to record large neural and behavioural data increases, there is growing interest in modeling neural dynamics during adaptive behaviors to probe neural representations. In particular, neural latent embeddings can reveal underlying correlates of behavior, yet, we lack non-linear techniques that can explicitly and flexibly leverage joint behavior and neural data. Here, we fill this gap with a novel method, CEBRA, that jointly uses behavioural and neural data in a hypothesis- or discovery-driven manner to produce consistent, high-performance latent spaces. We validate its accuracy and demonstrate our tool's utility for both calcium and electrophysiology datasets, across sensory and motor tasks, and in simple or complex behaviors across species. It allows for single and multi-session datasets to be leveraged for hypothesis testing or can be used label-free. Lastly, we show that CEBRA can be used for the mapping of space, uncovering complex kinematic features, and rapid, high-accuracy decoding of natural movies from visual cortex."> | ||
<meta property="og:image" content=""> | ||
|
||
<!-- Twitter --> | ||
<meta property="twitter:card" content="summary_large_image"> | ||
<meta property="twitter:url" content="https://cebra.ai/"> | ||
<meta property="twitter:title" content="Learnable latent embeddings for joint behavioural and neural analysis"> | ||
<meta property="twitter:description" content="Mapping behavioural actions to neural activity is a fundamental goal of neuroscience. As our ability to record large neural and behavioural data increases, there is growing interest in modeling neural dynamics during adaptive behaviors to probe neural representations. In particular, neural latent embeddings can reveal underlying correlates of behavior, yet, we lack non-linear techniques that can explicitly and flexibly leverage joint behavior and neural data. Here, we fill this gap with a novel method, CEBRA, that jointly uses behavioural and neural data in a hypothesis- or discovery-driven manner to produce consistent, high-performance latent spaces. We validate its accuracy and demonstrate our tool's utility for both calcium and electrophysiology datasets, across sensory and motor tasks, and in simple or complex behaviors across species. It allows for single and multi-session datasets to be leveraged for hypothesis testing or can be used label-free. Lastly, we show that CEBRA can be used for the mapping of space, uncovering complex kinematic features, and rapid, high-accuracy decoding of natural movies from visual cortex."> | ||
<meta property="twitter:image" content=""> | ||
|
||
<!-- Bootstrap CSS --> | ||
<link href="https://cdn.jsdelivr.net/npm/[email protected]/dist/css/bootstrap.min.css" rel="stylesheet" integrity="sha384-1BmE4kWBq78iYhFldvKuhfTAU6auU8tT94WrHftjDbrCEXSU1oBoqyl2QvZ6jIW3" crossorigin="anonymous"> | ||
|
||
<script id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script> | ||
|
||
<link href="https://fonts.googleapis.com/css2?family=IBM+Plex+Sans+Condensed&display=swap" rel="stylesheet"> | ||
<link href="https://fonts.googleapis.com/css2?family=IBM+Plex+Mono&display=swap" rel="stylesheet"> | ||
|
||
<link href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.13.1/css/all.min.css" rel="stylesheet"> | ||
|
||
<style> | ||
|
||
:root { | ||
--cebra-c: #1D29B8; | ||
--cebra-e: #6235E0; | ||
--cebra-b: #A045E8; | ||
--cebra-r: #BF1BB9; | ||
--cebra-a: #D4164F; | ||
} | ||
|
||
.main { | ||
font-family: 'IBM Plex Sans Condensed', sans-serif; | ||
color: gainsboro; | ||
} | ||
|
||
.container-fluid .col { | ||
width: 100%; | ||
padding-left: 0; | ||
padding-right: 0; | ||
} | ||
|
||
.code { | ||
font-family: 'IBM Plex Mono', monospace; | ||
} | ||
|
||
h3 { | ||
color: var(--cebra-r); | ||
} | ||
|
||
a { | ||
color: var(--cebra-r); | ||
font-family: 'IBM Plex Sans Condensed', sans-serif; | ||
} | ||
|
||
a:hover { | ||
color: var(--cebra-b); | ||
} | ||
|
||
.muted-link { | ||
color: #BF1BB9; | ||
} | ||
|
||
.paper-thumbnail { | ||
background-color: white; | ||
border-radius: 5%; | ||
} | ||
</style> | ||
|
||
<title>CEBRA</title> | ||
</head> | ||
|
||
<body style="background-color: rgb(0, 0, 0);"> | ||
<div class="container-fluid d-flex flex-column main"> | ||
<div class="row"> | ||
<div class="col-md-2"> | ||
</div> | ||
<div class="col-md-8" id="main-content"> | ||
<div class="row text-center my-5" id="#"> | ||
<h1>Rethinking pose estimation in crowds: overcoming the detection information bottleneck and ambiguity</h1> | ||
</div> | ||
|
||
<!-- Begin author list--> | ||
<div class="row text-center mb-4"> | ||
<div class="col-md-3 mb-4"></div> | ||
<div class="col-md-2 mb-4"> | ||
Mu Zhou*</br> | ||
EPFL | ||
</div> | ||
<div class="col-md-2 mb-4"> | ||
Lucas Stoffl*</br> | ||
EPFL | ||
</div> | ||
<div class="col-md-2 mb-4"> | ||
Mackenzie W. Mathis</br> | ||
EPFL | ||
<a href="https://www.mackenziemathislab.org/mackenziemathis" target="_blank"><i class="fas fa-link"></i></a> | ||
</div> | ||
<div class="col-md-2 mb-4"> | ||
Alexander Mathis</br> | ||
EPFL | ||
<a href="https://https://www.mathislab.org/" target="_blank"><i class="fas fa-link"></i></a> | ||
</div> | ||
</div> | ||
<!-- End author list--> | ||
|
||
<div class="row text-center"> | ||
<div class="col-md-2 mb-4"></div> | ||
<div class="col-sm-2 mb-2"></div> | ||
<div class="col-sm-2 mb-2"> | ||
<h4> | ||
<a href="https://github.com/amathislab/BUCTD" target="_blank"> <i class="fab fa-github"></i> | ||
Code | ||
</a> | ||
</h4> | ||
</div> | ||
<div class="col-sm-2 mb-2"> | ||
<h4> | ||
<a href="https://arxiv.org/abs/2306.07879" target="_blank"> | ||
<i class="fas fa-file-alt"></i> | ||
Paper | ||
</a> | ||
</h4> | ||
</div> | ||
</div> | ||
|
||
<div class="row pt-4"> | ||
<h3> | ||
<i class="fas fa-file"></i> | ||
Abstract | ||
</h3> | ||
</div> | ||
|
||
<div class="row"> | ||
|
||
</div> | ||
<div class="row"> | ||
<p> | ||
Frequent interactions between individuals are a | ||
fundamental challenge for pose estimation algorithms. | ||
Current pipelines either use an object detector together | ||
with a pose estimator (top-down approach), or localize | ||
all body parts first and then link them to predict the | ||
pose of individuals (bottom-up). Yet, when individuals | ||
closely interact, top-down methods are ill-defined due | ||
to overlapping individuals, and bottom-up methods often | ||
falsely infer connections to distant body parts. Thus, | ||
we propose a novel pipeline called bottom-up conditioned | ||
top-down pose estimation (BUCTD) that combines the | ||
strengths of bottom-up and top-down methods. Specifically, | ||
we propose to use a bottom-up model as the detector, | ||
which in addition to an estimated bounding box provides a | ||
pose proposal that is fed as condition to an attention-based | ||
top-down model. We demonstrate the performance and efficiency | ||
of our approach on animal and human pose estimation benchmarks. | ||
On CrowdPose and OCHuman, we outperform previous state-of-the-art | ||
models by a significant margin. We achieve 78.5 AP on CrowdPose | ||
and 47.2 AP on OCHuman, an improvement of 8.6% and 4.9% over | ||
the prior art, respectively. Furthermore, we show that our | ||
method has excellent performance on non-crowded datasets | ||
such as COCO, and strongly improves the performance on multi-animal | ||
benchmarks involving mice, fish and monkeys. | ||
</p> | ||
</div> | ||
<div class="row"> | ||
|
||
<div class="col-md-4 mb-3"> | ||
<video width="100%" autoplay loop muted preload="auto"> | ||
<source src="../source/video/buctd-iccv.mp4" type="video/mp4"> | ||
</video> | ||
</div> | ||
|
||
</div> | ||
|
||
<div class="row"> | ||
|
||
<div class="col-md-4 mb-3"> | ||
<img src="../source/gif/buctd-1.gif" alt="GIF"> | ||
</div> | ||
|
||
<div class="col-md-4 mb-3"> | ||
<img src="../source/gif/buctd-2.gif" alt="GIF"> | ||
</div> | ||
|
||
<div class="col-md-4 mb-3"> | ||
<img src="../source/gif/buctd-3.gif" alt="GIF"> | ||
</div> | ||
|
||
</div> | ||
|
||
|
||
<div class="row pt-4"> | ||
<h3> | ||
<i class="fas fa-graduation-cap"></i> | ||
BibTeX</h3> | ||
</div> | ||
<div class="row"> | ||
<p>Please cite our paper as follows:</p> | ||
</div> | ||
<div class="row justify-content-md-center"> | ||
<div class="col-sm-10 rounded p-3 m-2" style="background-color: rgb(20,20,20);"> | ||
<small class="code"> | ||
@misc{zhou2023iccv,<br/> | ||
title={Rethinking pose estimation in crowds: overcoming the detection information-bottleneck and ambiguity},<br/> | ||
author={Mu Zhou and Lucas Stoffl and Mackenzie W. Mathis and Alexander Mathis},<br/> | ||
year={2023},<br/> | ||
journal={IEEE/CVF International Conference on Computer Vision}<br/> | ||
} | ||
</small> | ||
</div> | ||
</div> | ||
|
||
<div class="row"> | ||
<small class="text-muted">Webpage designed using Bootstrap 5 and Fontawesome 5.</small> | ||
<a href="#" class="ml-auto"><i class="fas fa-sort-up"></i></a> | ||
</div> | ||
|
||
</div> | ||
</div> | ||
|
||
</div> | ||
</body> | ||
|
||
</html> |
Binary file not shown.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.