Archived updates
Hiring
Full-time RS and SWE: We are actively hiring full-time Research Scientist and Software Engineer to join us at several locations, including Menlo Park, Seattle and US remote.
Post-doc/visiting researchers: We have opening positions for post-doc and visiting researchers to conduct cutting edge research on object and scene understanding. Please contact me if you are interested.
Updates
22' 12/06, [Hiring finalized]. Research Internship Available at Meta Reality Labs. I am currently looking for a motivated research intern to work with me in 2023 summer. The broad research areas include 3D object and scene understanding and reconstruction, self- and language supervised representation learning, efficient on-device model architectures and 3D generative models. Please shoot me an email or click this link if you would like to apply.
22' 06/20: We are excited to release the source code of our CVPR 2022 work Unified Transformer Tracker!
22' 03/30: The full Continual Learning Challenge at CLVision workshop has officially started! EgoObjects dataset supports all 3 challenge tracks.
22' 03/29: Check out our CVPR 2022 work UTT, a Unified Transformer Tracker.
22' 03/17: Interested in Continual Learning? Do not miss this opportunity to learn more details of CLVision workshop in a live Q&A session.
22' 03/15, I am thrilled to announce EgoObjects, first large-scale object centric data set of egocentric videos. Our full release will cover over 40,000 videos (110+ hours) across 200 main object categories collected over 25 countries. Check out the demo release at Continual Learning Challenge in the CLVision workshop of CVPR 2022 [Meta Research Blog] [Twitter]
22' 01/20, 1 paper NASViT accepted by ICLR!
21' 07/22, 3 papers accepted by ICCV, including AutoTSNet, MViT and Visual Transformer! Congrats to all co-authors.
2021/04/05, the 3rd HACS Temporal Action Localization Challenge 2021 will be hosted at CVPR'21 ActivityNet Workshop. It includes two tracks below.
i) Fully supervised track
ii) Weakly supervised track.
2022 summer research internship at Meta AI Research: I am looking for research interns to work with me on object/scene understanding problems, including object category/instance/state detection, object segmentation, tracking, human-object interaction, 3D object pose estimation, and 3D scene reconstruction from video. Please shoot me an email if you are interested.
2021/03/15, WE ARE HIRING. We have research scientist/engineer positions at Facebook AI.
If you have background in object detection/segmentation/scene understanding/efficient models, and are excited about shipping research outcomes into products, please directly apply here or drop me an email.
2021/02/28, our latest work FP-NAS on neural architecture search has been provisionally accepted by CVPR 2021.
2020/11/23, two new papers FP-NAS and Visual Transformer are posted on Arxiv!
2020/06/12, I will be giving a talk on long-tailed recognition at Imperfect Data (LID) Workshop during CVPR 2020.
2020/06/12, I am co-organizing HACS Temporal Action Localization Challenge 2020 at CVPR'20 International Challenge on Activity Recognition Workshop. The final challenge results can be found here.
2020/03/19, the code of our work on long-tailed recognition "Decoupling Representation and Classifier for Long-Tailed Recognition" is released !
2019/12/19, our work on long-tailed recognition is accepted by ICLR !
2019/12/04, ClassyVision.ai is officially online and open sourced ! ClassyVision is an end-to-end PyTorch framework for image and video classification. It is easy to write and launch distributed training jobs to train image and video models !
09/25/2019, Internship Opportunity: I plan to take a Ph.D research intern to work with me on video understanding during 2020 summer. If you are currently working in this area and are interested in this internship, please send me an email with your resume, and a short description on your research experience. Also check the full job description.
07/30/2019, code for DMC-Net, Global Reasoning Networks and Octave-Conv are released
07/18/2019, I am co-organizing the "Workshop on Multi-modal Video Analysis and Moments in Time Challenge" at ICCV 2019. We invite you to participate in two challenges
A) Multi-modal Multi-label Action Detection Challenge using Moments in Time dataset
B) Temporal Action Localization Challenge using HACS Segments Dataset
Challenge Portal is open now!
04/10/2019, check our new work on reducing spatial redundancy by proposing a novel Octave Convolution operation
02/25/2019. Two papers [DMC-Net] and [Global Reasoning Network] are accepted by CVPR 2019.
02/01/2019. HACS: Human Action Clips and Segments (previously known as SLAC) dataset is officially released [Github]. It has two parts. 1) HACS Clips consists of 1.55M 2-second clips from 520K videos annotated for action classification. 2) HACS Segments contains 139K manually annotated segments from 50K untrimmed videos for action detection.
01/15/2019, internship opportunities: I am looking for a strong research intern to work with me in 2019 summer. The research topic will be around video understanding. Please shoot me an email with CV, mention your research interests and what project you would like to work on if you are interested.
01/12/2019, check out our new work DMC-Net, where we propose a lightweight generator network to reduce noises in motion vectors and capture fine motion details. It achieves a more Discriminative Motion Cue (DMC) representation for compressed video action recognition.
12/01/2018, our new work Global Reasoning Networks, which proposes a lightweight and end-to-end trainable unit for improving image classification/semantic segmentation and video classification, is deposited on Arxiv.
09/28/2018, as part of Facebook AI Computer Vision talks, I will give a presentation at Siebel Center 2405, on Oct 12, 12pm, and have 1:1: meeting with students afterwards. Please RSVP below.
12/01/2017, our work on action classification and localization is online.
11/16/2017, Facebook Computer Vision team have a few opening positions of 2018 summer research internship on video understanding. Please send me an email if you are interested.
05/16/2017, new paper on FCN-based video stylization accepted by IEEE TIP.
10/02/2017, internship opportunities: I am happy to work with passionate and talented PhD students on research projects. Opening positions in Spring and Summer 2017 are available now. To apply, send me an email with your resume, mention your research interests and what project you would like to work on.
Call for paper. The 1st Workshop on Large Scale Computer Vision Systems will be held during NIPS, 2016.
The code of hybrid network for semantic segmentation is released.
05/24/2016, one paper on learning concept taxonomies is accepted by ACL 2016.
03/31/2016, our hybrid network for semantic segmentation is online.
02/19/2016, Ph.D thesis defense is completed.
12/10/2015, I will join Facebook as a Research Scientist in 2016 summer.
08/29/2015, HD-CNN paper is accepted by ICCV 2015.
05/18/2015, summer internship@Google Brain starts
05/17/2015, code and dataset for the TOG paper 'Automatic Photo Adjustment Using Deep Neural Networks' are released
05/17/2015, HD-CNN code is released. Project page is online.
02/26/2015, our work 'Automatic Photo Adjustment Using Deep Neural Networks' is accepted by ACM TOG
Awards and Honors
CVPR Outstanding Reviewer Award, 2020
Chinese Government Award for Outstanding Self-Financed Students Abroad, 2015
Doctoral Consortium with Travel Grant, ICCV 2015
Tencent Technology Scholarship (Top 1%), Highest Honor for Graduate Students in Zhejiang University, 2009
Qimonda Technology Scholarship (Top 5%), Zhejiang University, 2008