Laura Sevilla-Lara

Timeline

1. 2010–2015

   PhD at UMass Amherst

   After a computer engineering degree at the University of Granada and a master's at Brown University, she studies for a computer science PhD at the University of Massachusetts Amherst from 2010 to 2015.

2. 2015–2017

   Postdoc at Max Planck

   From 2015 to 2017, she is a postdoctoral researcher in the Perceiving Systems group at the Max Planck Institute for Intelligent Systems. In March 2016, she coauthors Optical Flow with Semantic Segmentation and Localized Layers with Deqing Sun, Varun Jampani and Michael J. Black, a semantic flow method with the lowest error of any published monocular method on the KITTI-2015 flow benchmark. In December 2017, she coauthors On the Integration of Optical Flow and Action Recognition with Yiyi Liao, Fatma Güney, Varun Jampani, Andreas Geiger and Michael J. Black, which examines why optical flow helps action recognition.

3. 2017–2018

   Lecturer at Edinburgh

   From August 2017, she is a visiting researcher at Facebook Research in Menlo Park, California. In 2018, she joins the School of Informatics at the University of Edinburgh as a lecturer in computer vision.

4. 2019

   Only Time Can Tell

   In 2019, she coauthors DMC-Net, a compressed-video action recognition network that runs two orders of magnitude faster than optical flow at inference time, and FASTER, a recurrent network that aggregates predictions from models of different complexities. She also writes Only Time Can Tell with Shengxin Zha, Zhicheng Yan, Vedanuj Goswami, Matt Feiszli and Lorenzo Torresani, which gathers the action classes that need temporal information into a temporal dataset.

5. 2020–2022

   Google awards and BMVC chair

   She receives a Google Faculty Award in 2020. In 2021, she is one of six programme chairs of the 32nd British Machine Vision Conference, held in November 2021. In 2022, she receives a Google Scholar Award in the area of Machine Perception from the Research Scholar Program for early-career faculty, for the project Sparse Temporal Representations for Video Understanding.

6. 2023–2024

   Affordances for robot manipulation

   In November 2023, she coauthors One-Shot Open Affordance Learning with Foundation Models with Gen Li, Deqing Sun and Varun Jampani, a vision-language framework that outperforms state-of-the-art models with less than 1% of the full training data. In August 2024, she coauthors Learning Precise Affordances from Egocentric Videos for Robotic Manipulation, a system that reaches 77.1% successful grasping among 179 trials.

7. Since 2024

   Reader at Edinburgh

   She becomes a Reader (Associate Professor) in Edinburgh's School of Informatics in August 2024. In 2024, she gives the keynote Frontiers of Video Understanding at the 35th British Machine Vision Conference. She is also an ELLIS Scholar.