Tightly Connecting Vision and Language

Tightly Connecting Vision and Language

Channel:
Microsoft Research
Subscribers:
344,000
Published on ● Video Link: https://www.youtube.com/watch?v=3D7XwXDCZ30


Duration: 1:07:38
1,098 views
31

Remarkable progress has been made at the intersection of vision and language. While showing great promise, current vision and language models may only weakly "connect" the two modalities and often fail in the wild. In this talk, I will present our recent efforts aiming to bridge this gap along two dimensions: informativeness and controllability. In particular, I will describe how we can leverage large-scale datasets, including our recently-released CC12M and Localized Narratives, to benefit existing vision-and-language tasks as well as to enable new applications.

Speaker: Soravit (Beer) Changpinyo, Google

MSR Deep Learning team: https://www.microsoft.com/en-us/research/group/deep-learning-group/



Other Videos By Microsoft Research


2021-10-05Women of Color and the Digital Labor of Repair
2021-10-01Fake It Till You Make It: Face Analysis In The Wild Using Synthetic Data Alone
2021-09-23ALIGN: Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision
2021-09-23Zero-Shot Detection via Vision and Language Knowledge Distillation
2021-09-17Three Explorations on Pre-Training: an Analysis, an Approach, and an Architecture
2021-09-16Visual Recognition beyond Appearances, and its Robotic Applications
2021-09-16A Truly Unbiased Model
2021-09-16Visual question answering & reasoning over vision & language: Beyond limits of statistical learning?
2021-09-15MDETR: Modulated Detection for End-to-End Multi-Modal Understanding
2021-09-15Learning Commonsense Understanding through Language and Vision
2021-09-15Tightly Connecting Vision and Language
2021-09-15Learning from Unlabeled Videos for Recognition, Prediction, and Control
2021-09-15Grounded Visual Generation
2021-08-25The New Jim Code: Reimagining the Default Settings of Technology & Society
2021-08-19A mechatronic shape display based on auxetic materials
2021-08-16Dependable IoT- Making data from IoT devices dependable and trustworthy for good decision making
2021-08-11Lookout System: National Television Commercial (1998)
2021-08-06Create human-centered AI with the Human-AI eXperience (HAX) Toolkit webinar
2021-08-04Computing Technology as Racial Infrastructure: A History of the Present & Blueprint for Black Future
2021-07-27Urban Air Chicago
2021-07-09The Vanishing Indian Speaks Back: Race, Genomics, and Indigenous Rights