Language in Vision: Visual Reasoning, Visual Navigation, Grounded Translation

Add to your list(s) Download to your calendar using vCal

• Mateusz Malinowski (DeepMind)
• Friday 19 February 2021, 17:00-18:00
• https://cern.zoom.us/j/67127006104?pwd=SkwxUWN4Zm1ST3BEKzJTSGwycU53Zz09.

If you have a question about this talk, please contact srj38.

With deep learning and large-scale datasets, we can work on increasingly more ambitious computer vision problems and multimodal settings. In this presentation, I will talk about three computer vision problems where natural or synthetic language has played a fundamental role. Those problems are: 1) visual reasoning that takes the form of question answering about visual scenes, 2) visual navigation where the language provides instructions for the agent to follow, 3) grounded translation where, for a change, vision may turn out to be important for training translation models.

This talk is part of the CuAI (Cambridge University Artificial Intelligence Society) series.

This talk is included in these lists:

• CuAI (Cambridge University Artificial Intelligence Society)
• https://cern.zoom.us/j/67127006104?pwd=SkwxUWN4Zm1ST3BEKzJTSGwycU53Zz09

Note that ex-directory lists are not shown.