The 14th European Conference on Computer Vision – Amsterdam, The Netherlands
ECCV awards 2016: "Unsupervised Domain Adaptation with Regularized Domain Instance Denoising", Gabriela Csurka, Boris Chidlovskii, Stephane Clinchant and Sofia Michel got one of the two best paper awards at the ECCV TASK-CV workshop. And Diane Larlus got a best reviewer award.
8th October, 2016:
New directions in saliency research: Developments in architecture, datasets and evaluation workshop; Naila Murray, invited talk: "Deep networks for eye fixation prediction"
Cesar De Souza and Adrien Gaidon: Sympathy for the Details: Dense Trajectories and Hybrid Classification Architectures for Action Recognition, co-authored with Eleonora Vig and Antonio Lopez.
8th October, 2016:
Geometry Meets Deep Learning workshop: ”Learning the semantic structure of objects from Web supervision”, David Novotny, Diane Larlus and A. Vedaldi. This Paper will also be presented (poster session): October 10th, 2016, ECCV Workshop on Web-scale Vision and Social Media
15th October 2016:
ECCV invited tutorial: Florent Perronnin: "Output embedding for large-scale visual recognition"
Abstract: The focus of the computer vision community has long been on input embedding: how to transform an image or a video into a suitable descriptor? During this tutorial, we will consider the output embedding problem: how to embed classes in a Euclidean space? We will see that output embedding is a must for large-scale visual recognition as it enables parameter sharing: this yields classifiers which are more accurate when training data is scarce (including zero-shot recognition) and which are faster to train and evaluate. We will provide a taxonomy of output embeddings and focus on embeddings based on a priori information (eg class hierarchies, attributes or textual descriptions). We will also explain how to measure the compatibility between input and output embeddings, starting from simple least-square regression and going all the way to deep learning. Finally, we will show recent successful applications of output embedding to web-scale visual classification.