recent news

  1. Four new papers to be presented at CVPR 2022:

  2. Learning To Recognize Procedural Activities with Distant Supervision,
    with Xudong Lin, Fabio Petroni, Gedas Bertasius, Marcus Rohrbach, and Shih-Fu Chang. 

  3. Deformable Video Transformer,
    with Jue Wang.

  4. Long-Short Temporal Contrastive Learning of Video Transformers,
    with Jue Wang, Gedas Bertasius, and Du Tran.

  5. Ego4D: Around the World in 3,000 Hours of Egocentric Video.
    Project page and dataset available here.

  1. New paper presented at AAAI 2022:

  2. Label Hallucination for Few-Shot Classification,
    with Yiren Jian.

  1. New article published in IEEE Transactions on Pattern Analysis and Machine Intelligence:

  2. Generalized Few-Shot Video Classification with Video Retrieval and Feature Generation,
    with Yongqin Xian, Bruno Korbar, Matthijs Douze, Bernt Schiele, and Zeynep Akata.

  1. I gave three keynotes at CVPR 2021 workshops:

  2. "Video Understanding with Language Models," EPIC@CVPR2021, The Eight International Workshop on Egocentric Perception, Interaction and Computing.
    [video link]

  3. "Vision using Sight... but also Sound and Speech," MULA, The Fourth Multimodal Learning and Applications Workshop. [video link]

  4. "Space-Time Models for Segmentation, Tracking and Recognition in Video," RVSU, Robust Video Scene Understanding Workshop. [video link]

  1. A Facebook AI blog article describing our new research on video understanding. Code and pretrained models are available here.

  1. Two new papers to be presented at ICML 2021:

  2. Is Space-Time Attention All You Need for Video Understanding?,
    with Gedas Bertasius, and Heng Wang. 

  3. Slot Machines: Discovering Winning Combinations of Random Weights in Neural Networks ,
    with Maxwell Aladago.

  1. Two new papers presented at CVPR 2021:

  2. VX2TEXT: End-to-End Learning of Video-Based Text Generation From Multimodal Inputs,
    with Xudong Lin, Gedas Bertasius, Jue Wang, Shih-Fu Chang, and Devi Parikh. 

  3. Beyond Short Clips: End-to-End Video-Level Learning with Collaborative Memories,
    with Xitong Yang, Haoqi Fan, Larry Davis, and Heng Wang.

research overview

My research interests are in computer vision and machine learning. My current work is primarily focused on learning representations for image and video understanding.

previous affiliations

  1. Dartmouth, Computer Science

  2. Fulbright U.S. Scholar at Ashesi University in Ghana.

  3. Microsoft Research Cambridge, Machine Learning and Perception

  4. Riya/

  5. New York University, Computer Science

  6. Stanford University, Computer Science

  7. DigitalPersona

  8. IRST

  9. University of Milan, Computer Science

Lorenzo Torresani

Research Lead

Facebook AI Research (FAIR), Meta

Email / Google Scholar