ICCV 2017 Tutorial on

Instance-level Visual Recognition

Venice, Italy
Oct 22nd, 2017


Speakers

Georgia Gkioxari
FAIR
Ross Girshick
FAIR
Kaiming He
FAIR
Justin Johnson
Stanford
Jifeng Dai
MSRA
Piotr Dollàr
FAIR


Overview

The purpose of this tutorial is to discuss the recent advances on instance-level recognition. We will cover in detail the most recent work on object detection, instance segmentation and human pose prediction from a single image. Going beyond single images, we will show the most recent progress in video object understanding. Finally, we will present how instance-level recognition can benefit tasks such as human-object interactions, visual relationships and instance-level reasoning.


Schedule

14:00 - 14:10 . Welcome and Overview. slides

14:10 - 14:50 . Talk 1: The Generalized R-CNN Framework for Object Detection by Ross Girshick slides

14:50 - 15:20 . Talk 2: Mask R-CNN: A Perspective on Equivariance by Kaiming He slides

15:20 - 15:50 . Talk 3: Recognizing Human-Object Interactions by Georgia Gkioxari slides

15:50 - 16:30 . Break.

16:30 - 17:00 . Talk 4: Flow-based Video Recognition by Jifeng Dai slides

17:00 - 17:30 . Talk 5: Beyond Object Instances by Justin Johnson slides

17:30 - 17:45 . Closing Remarks


Contact: Georgia Gkioxari