What is Recluse

ReCluSe is a self-help reading group where everybody is supposed to contribute with questions, answers, comments and observations. See Instructions for further details.

Where and When

ReCluSe currently takes place every Wednesday in room ETF/C109 at 15:00 (no academic quarter). Meetings might be suspended or moved in some circumstances (e.g. holidays, deadlines, time overlaps). You will be reminded of meetings through the mailing list.

Mailing List

If you want to participate to ReCluSe, please send an email to ReCluSe-request@list.ee.ethz.ch with subject "subscribe" and an empty body.

ReCluSe Current Status

In 2013 ReCluSe started again with papers from major computer vision conferences. The list and prelimary schedule of paper assignments:
Date Title Authors Conference Summary
13.02.2014 THU 15:00 (new) Struck: Structured Output Tracking with Kernels S. Hare, A. Saffari, P. Torr ICCV 2011 -
29.01.2015 WED 15:00 Co-Hierarchical Analysis of Shape Structures O. van Kaick, K. Xu, H. Zhang, Y. Wang, S. Sun, A. Shamir, D. Cohen-Or SIGGRAPH 2013 -
15.01.2014 WED 15:00 A Connection between Partial Symmetry and Inverse Procedural Modeling M. Bokeloh, M. Wand, H.-P. Seidel SIGGRAPH 2010 -
19.11.2013 WED 15:00 Dense Reconstruction Using 3D Object Shape Priors A. Dame, V. Prisacariu, C. Ren. CVPR 2013 Summary
09.10.2013 WED 15:00 Video Object Segmentation through Spatially Accurate and Temporally Dense Extraction of Primary Object Regions D. Zhang, O. Javed, M. Shah CVPR 2013 Summary
25.09.2013 WED 10:00 Diffusion Processes for Retrieval Revisited M. Donoser and H. Bischof CVPR 2013 Summary
11.09.2013 WED 15:00 Large Displacement Optical Flow from Nearest Neighbor Fields Z. Chen, H. Jin, Z. Lin, S. Cohen, Y. Wu CVPR 2013 Summary
14.08.2013 WED 15:00 2.5 Dual contouring with global regularities Q-Y. Zhou and U. Neumann CVPR 2012 Summary
31.07.2013 WED 15:00 The Visual Turing Test for scene reconstruction Q. Shan, R. Adams, B. Curless, Y. Furukawa, S. Seitz 3DV 2013 (was 3DPVT) Summary
17.07.2013 WED 15:00 What makes Paris look like Paris? C. Doersch, S. Singh, A. Gupta, J. Sivic, A. Efros SIGGRAPH 2012 Summary
26.06.2013 WED 15:00 Affinity propagation B. Frey and D. Dueck Science 2007 Summary
12.06.2013 WED 15:00 Accelerated Hypothesis Generation for Multi-structure Robust Fitting T.-J. Chin, J. Yu, D. Suter ECCV 2010 Summary
16.05.2013 THU 15:00 Joint Shape Segmentation with Linear Programming Q-X. Huang, V. Koltun, L. Guibas SIGGRAPH 2011 Summary
02.05.2013 THU 15:00 Scene Shape from Texture of Objects N. Payet and S. Todorovic CVPR 2011 Summary
24.04.2013 WED 16:00 Matching local self-similarities across images and videos E. Shechtman and M. Irani CVPR 2007 Summary
17.04.2013 16:00 Discrete-continuous optimization for large-scale structure from motion D. Crandall, A. Owens, N. Snavely, D. Huttenlocher CVPR 2011 Summary
20.03.2013 16:00 Shape Bolzman Machine: A Strong Model of Shape S. Eslami, N. Heess, J. Winn CVPR 2012 Summary
13.03.2013 16:00 Robust Multiple Structure Estimation - JLinkage R. Toldo and A. Fusiello ECCV 2008 Summary
06.03.2013 16:00 Reconstructing the World's Museums J. Xiao and Y. Furukawa ECCV 2012 Summary
27.02.2013 16:00 Diverse M-best Solutions in MRFs D. Batra, P. Yadollahpour, A. Guzman-Rivera, G. Shakhnarovich ECCV 2012 Summary
20.02.2013 16:00 Structural Approach for Building Reconstruction from a DSM F. Lafarge, X. Descombes, J. Zerubia, M. P-Deseilligny PAMI 2010 Summary
13.02.2013 16:00 Manhattan-World Stereo Y. Furukawa, B. Curless, S. Seitz, R. Szeliski CVPR 2009 Summary
06.02.2013 16:00 Summarizing Visual Data using Bidirectional Similarity D. Simakov, Y. Caspi, E. Shechtman, M. Irani CVPR 2008 Summary
31.01.2013 16:00 Texture Synthesis by Non-parametric Sampling A. Efros and T. Leung ICCV 1999 Summary
23.01.2013 16:00 Diagnosing Error in Object Detectors D. Hoiem, Y. Chodpathumwan, Q. Da ECCV 2012 Summary


A paper is selected by the ReCluSe participants. Every member of ReCluSe can propose a paper to read. When one paper session has been completed, a new voting takes place and a new set of papers is selected.

ReCluSe IS NOT a lecture. Participants are supposed to read the paper in the days before the meeting. At the meeting we won't read the paper, but go through it page by page, and stop whenever there is a question from a member of the group. The idea is that when someone did not understand a point, he stops the flow and asks a question. Chances are, that someone else understood that point, and can answer. Ideally, this will cause the collective understanding to be the 'set union' of the individual's understanding.

Attendance is entirely optional. It can happen that a particular paper that is chosen is of no interest to you. However, you should vote for a paper only if you think you can make time to attend the meetings in case your paper is selected.

Completed Sessions

This is the list of ReCluSe sessions that we completed during past recluse meetings: Will be updated soon.