Fabian Caba Heilbron

I am a Research Scientist at Adobe Research. I recieved my Ph.D. from KAUST under the supervision of Professor Bernard Ghanem. I was fortunate to work with Juan Carlos Niebles during my master studies at Universidad del Norte.

My research interests span computer vision and machine learning. Specifically, my research focuses in the developments of models to recognize and detect human activities from videos. Some of my previous efforts include:

  • gathering large-scale video datasets
  • developing efficient scanning methods for video analysis
  • training models for temporal activity localization

If you want to reach out, please contact me at: fabian.caba [at] kaust.edu.sa

Professional Experience

ActivityNet Challenge [2016-Present]: I am one of the leading forces behind organizing the ActivityNet Challenge. I built the infrastructure that allows multiple researchers to test-bed their methods. The challenge has attracted a large number of researchers, and it has been sponsored by several companies including DeepMind, Facebook, and Google AI.

Mantis AI [2017-Present]: I am a co-founder of Mantis AI. Mantis crafts ground-breaking research in deep-learning to perform activity detection and video summarization for content-aware ad placement.

IVUL [2015-2019]: I was a Research Assistant at IVUL. My work aimed to develop models for human activity understanding. Some of my efforts were published at top-tier computer vision conferences (see Publications).

Adobe Research [Summer 2017, Summer 2018]: I was an intern for two consecutive summers at Adobe Research. During these periods I developed: (i) an active learning method for action detection; (ii) a navigation system for video editing purposes.

Publications


What do I Annotate Next? An Empirical Study of
Active Learning for Action Localization

Fabian Caba Heilbron, Joon-Young Lee, Hailin Jin, Bernard Ghanem
Munich, ECCV 2018. - [Paper] [Project page]


Action Search: Spotting Targets in Videos and Its Application to
Temporal Action Localization

Humam Alwassel*, Fabian Caba Heilbron*, Bernard Ghanem; * := equal contribution
Munich, ECCV 2018. - [Paper] [Project page] [Video]

Diagnosing Error in Temporal Action Detectors
Humam Alwassel*, Fabian Caba Heilbron*, Victor Escorcia*, Bernard Ghanem
* := equal contribution
Munich, ECCV 2018. - [Paper] [Project page] [Code (Github)] [Video]


SCC: Semantic Context Cascade for Efficient Action Detection
Fabian Caba Heilbron, Wayner Barrios, Victor Escorcia, Bernard Ghanem
Hawaii, CVPR 2017. - [Paper] [Poster]


DAPs: Deep Action Proposals for Action Understanding
Victor Escorcia, Fabian Caba Heilbron, Juan Carlos Niebles, Bernard Ghanem
Amsterdam, ECCV 2016. - [Paper] [Code (Github)] [Poster]


Fast Temporal Activity Proposals for
Efficient Detection of Human Actions in Untrimmed Videos

Fabian Caba Heilbron, Juan Carlos Niebles, Bernard Ghanem
Las Vegas, CVPR 2016. - [Paper] [Project page] [Code (Github)] [Demo]

ActivityNet: A Large-Scale Video Benchmark for Human Activity Understanding
Fabian Caba Heilbron, Victor Escorcia, Bernard Ghanem, Juan Carlos Niebles
Boston, CVPR 2015. - [Paper] [Project page] [Code (Github)] [Demo]

Robust Manhattan Frame Estimation from a Single RGB-D Image
Bernard Ghanem, Ali Thabet, Juan Carlos Niebles, Fabian Caba Heilbron
Boston, CVPR 2015. - [Paper] [Project page] [Code] [Data]


Camera Motion and Surrounding Scene Appearance as Context for Action Recognition
Fabian Caba Heilbron, Ali Thabet, Juan Carlos Niebles, Bernard Ghanem
Singapore, ACCV 2014. - [Paper] [Project page]

Collecting and Annotating Human Activities in Web Videos
Fabian Caba Heilbron, Juan Carlos Niebles
Glasgow, ICMR 2014. - [Paper] [Code (Github)]

In God we trust. All others must bring data.
W. Edwards Deming