AlexandreAlahiAlexandreAlahiAlexandreAlahiAlexandreAlahiAlexandreAlahiAlexandreAlahiAlexandreAlahiAlexandreAlahiAlexandreAlahi

Research overview

Although the progress in collecting visual data brought unprecedented opportunities, it also raises the following fundamental challenge:

How to extract meaningful information from an overwhelming volume of noisy multi-modal visual data and learn from it?

The captured signals are high dimensional and high entropy; the meaningful information is unsegmented, potentially in part due to occlusions, or even ambiguous due to the nature of the modality (e.g., missing 3D information in the RGB data). Moreover, the causality behind the data (e.g., the set of route choices a person takes while walking in a crowded space) is due to both observable cues (the dynamic and static content of the scene) as well as non-observable cues (e.g., the intention of the individuals or social conventions). Mathematically, it leads to ill-posed, non-convex, intractable problems.

My work in addressing these challenges brings together computer vision, data/signal processing, machine learning, statistics, distributed systems, and insights from neuroscience and psychology.

In this page, I present my effort in addressing the above challenges with a focus on human behavior understanding at scale:
From 1) Sensing, 2) Perception, 3) Learning, 4) Communication, to 5) Action.

In a nutshell, I have proposed sparsity promoting priors, discrete combinatorial optimization, attention models, and new recurrent neural network architectures to learn the causalities behind the world we observe.

 

    I - Sensing Human Behavior at Scale

Graph matching

RGB-W: When Vision Meets Wireless
Alexandre Alahi, Albert Haque, and Li Fei-Fei.
IEEE International Conference on Computer Vision (ICCV), 2015.
[pdf | project page]

People tracking


Socially-aware large-scale crowd forecasting
Alexandre Alahi, Vignesh Ramanathan, and Li Fei-Fei.
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014
.
(Oral
, acceptance rate: 5.8%)
.
[pdf ]



    II- Real-time Perception

II- A) Low-level Descriptor

freak

Biologically Inspired Keypoints
Alexandre Alahi, Georges Goetz, and Emmanuel D’Angelo.
Book
 on Biologically Inspired Computer Vision: Fundamentals and Applications, 2015.
[pdf | project page]

freak
From bits to images: Inversion of local binary descriptors
Emmanuel D'Angelo, Jacques L, Alexandre Alahi, Pierre Vandergheynst.
IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2014.
[pdf | project page]
freak Beyond bits: Reconstructing images from local binary descriptors
Emmanuel D'Angelo, Alexandre Alahi, and Pierre Vandergheynst.  
IEEE International Conference on Pattern Recognition (ICPR), 2012
.
[pdf | project page]

freak

FREAK: Fast Retina Keypoint
Alexandre Alahi, Raphael Ortiz, and Pierre Vandergheynst.  
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012
.
(Open Source Award Winner)
[pdf | project page]

Graph matching Foreground silhouette extraction robust to sudden changes of background appearance
Alexandre Alahi, Luigi Bagnato, Damien Matti, and Pierre Vandergheynst.
IEEE International Conference on Image Processing (ICIP), 2012.
[pdf]


II- B) Human Detection in 3D (Locating Humans)

Graph matching RGB-W: When Vision Meets Wireless
Alexandre Alahi, Albert Haque, and Li Fei-Fei.  
IEEE International Conference on Computer Vision (ICCV), 2015.
[pdf | project page]
Graph matching

Robust real-time pedestrians detection in urban environments with low-resolution camera
Alexandre Alahi, Michel Bierlaire, and Pierre Vandergheynst.
Transportation research part C: emerging technologies, 2014.
[pdf]

Graph matching Scoop: A real-time sparsity driven people localization algorithm
Mohammad Golbabaee, Alexandre Alahi, and Pierre Vandergheynst.
 Journal of mathematical imaging and vision, 2014.
[pdf | project page]
Graph matching Method and system for automatic objects localization
Alexandre Alahi, Mohammad Golbabaee, and Pierre Vandergheynst.
U.S. Patent No. 8,749,630. 10 Jun. 2014.

[pdf | project page]
Graph matching Sparsity driven people localization with a heterogeneous network of cameras
Alexandre Alahi, Laurent Jacques, Yannick Boursier, and Pierre Vandergheynst.  
Journal of Mathematical Imaging and Vision, 2011.
[pdf | project page]
Graph matching Sparsity-driven people localization algorithm: Evaluation in crowded scenes environments
Alexandre Alahi, Laurent Jacques, Yannick Boursier, and Pierre Vandergheynst.  
Performance Evaluation of Tracking and Surveillance (PETS-Winter), 2009.

[pdf | project page]
Graph matching

Sport players detection and tracking with a mixed network of planar and omnidirectional cameras
Alexandre Alahi, Yannick Boursier, Laurent Jacques, and Pierre Vandergheynst.  
ACM/IEEE International Conference on Distributed Smart Cameras (ICDSC), 2009
.
(Challenge price winner)
[pdf | project page]

Graph matching A sparsity constrained inverse problem to locate people in a network of cameras 
Alexandre Alahi, Yannick Boursier, Laurent Jacques, and Pierre Vandergheynst. 
IEEE International Conference on Digital Signal Processing, 2009.

[pdf | project page]

II- C) Object Tracking

freak Coming soon: End-to-end learning to track any object: from ants to bicyclists...
freak Recurrent Attention Models for Depth-Based Person Identification
Albert Haque, Alexandre Alahi, and Li Fei-Fei.
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
[pdf | project page]
freak
Learning to Track: Online Multi-Object Tracking by Decision Making
Yu Xiang, Alexandre Alahi, and Silvio Savarese.  
IEEE International Conference on Computer Vision (ICCV), 2015.
(Oral, acceptance rate: 3.3%)
[pdf | project page]
freak Cascade of descriptors to detect and track objects across any network of cameras 
Alexandre Alahi, Pierre Vandergheynst, Michel Bierlaire, and Murat Kunt.
Journal on Computer Vision and Image Understanding, 2010.
[pdf | project page]
freak

Object detection and matching with mobile cameras collaborating with fixed cameras
Alexandre Alahi, Michel Bierlaire, and Murat Kunt. 
ECCV workshop on Multi-camera and Multi-modal Sensor Fusion
, 2008.
[pdf | project page]

freak A master-slave approach for object detection and matching with fixed and mobile cameras
Alexandre Alahi, David Marimon, Michel Bierlaire, and Murat Kunt.
IEEE International Conference on Image Processing (ICIP), 2008.
[pdf | project page]
freak Object detection and matching in a mixed network of fixed and mobile cameras
Alexandre Alahi, Pierre Vandergheynst, Michel Bierlaire, and Murat Kunt.
ACM Multimedia workshop on analysis and retrieval of events/actions, 2008.
[pdf | project page]

II- D) Human Pose Estimation

People tracking


Towards Viewpoint Invariant 3D Human Pose Estimation
Albert Haque, Boya Peng*, Zelun Luo*, Alexandre Alahi, Serena Yeung, Li Fei-Fei.
European Conference on Computer Vision (ECCV), 2016.
[pdf | project page]



    III- Learning to Predict Human Dynamics

People tracking

Learning Social Etiquette
Alexandre Robicquet, Amir Sadeghian, Alexandre Alahi, Silvio Savarese.
European Conference on Computer Vision (ECCV), 2016.
[pdf | project page]

People tracking

Knowledge Transfer for Scene-specific Motion Prediction
Lamberto Ballan, Francesco Castaldo, Alexandre Alahi, Francesco Palmieri, Silvio Savarese.
European Conference on Computer Vision (ECCV), 2016.
[pdf | project page]

People tracking

Social LSTM: Human Trajectory Prediction in Crowded Spaces
Alexandre Alahi*, Kratarth Goel*, Vignesh Ramanathan, Alexandre Robicquet, Li Fei-Fei, Silvio Savarese.
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. (* equal contribution)
(Spotlight oral)

[pdf | project page]

People tracking Socially-aware large-scale crowd forecasting
Alexandre Alahi, Vignesh Ramanathan, and Li Fei-Fei.  
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014
.
(Oral
, acceptance rate: 5.8%)
.
[pdf | project page]

    IV- Communication/Visualization

People tracking

Perceptual Losses for Real-time Style Transfer and Single Image Super-Resolution
Justin Johnson, Alexandre Alahi, and Li Fei-Fei.
European Conference on Computer Vision (ECCV), 2016.
[pdf | project page]

People tracking


Activity Understanding from Multi-modal Visual and Physiological Signals
Katsuyuki Nakamura, Serena Yeung, Alexandre Alahi, and Li Fei-Fei.
Coming soon
[pdf | project page]

Graph matching System and method for media library navigation and recommendation
Alexandre Alahi, Pierre Vandergheynst, and Kirell Benzi. 
No. EPFL-PATENT-211599. 2014.

[pdf | project page]
Graph matching Stream carving: an adaptive seam carving algorithm
Daniel Domingues, Alexandre Alahi, and Pierre Vandergheynst.  
IEEE International Conference on Image Processing (ICIP), 2010.

[pdf | project page]
People tracking

System and method for measuring performances of surveillance systems
Ali Azarbayejani, Alexandre Alahi, and Murat Erdem.
U.S. Patent No. 7,415,385. 19 Aug. 2008.
[pdf | project page]

   

    V- Action

People tracking

Coming soon: our social robot

[pdf | project page]

People tracking

Vision-Based Hand Hygiene Monitoring in Hospitals
Yeung S, Alahi A, Haque A, Luo Z, Peng B, Singh A, Platchek T, Milstein A, Fei-Fei L. .
- In Proceedings of the American Medical Informatics Association (AMIA), 2016
- In Proceedings of the NIPS Workshop on Machine Learning in Healthcare, 2015

[pdf | project page]


    and more to come soon!

People tracking Coming soon: learning human navigation in complex campus scenes...

[pdf | project page]
People tracking

Coming soon: learning camera network topology

[pdf | project page]


 


top

© alahi {at} stanford.edu
updated: October 2016

Follow me on

Alahi on ScholarAlahi on LinkedinAlahi on Twitter