In addition to the main four day program of presentations, interactive sessions, plenary talks, demos, exhibitions, and social functions, CVPR 2016 has a number of colocated events, including 29 workshops and 22 tutorials. As the field of artificial intelligence has become a major player in the technology world, this year's CVPR has made history in a number of exciting ways.
First and foremost, we received a record 2145 valid submissions to the main conference, of which 1865 were fully reviewed (the others were either administratively rejected for technical or ethical reasons or withdrawn before review). To select papers from these submissions, we invited 72 researchers to act as Area Chairs (ACs). ACs were selected to provide a broad range of expertise, to balance junior and senior members, and to represent a variety of geographical locations. Additionally, we recruited a team of 1141 experienced reviewers from the broader computer vision and pattern recognition community. The original list of reviewers was augmented with reviewers recommended by the ACs to add expertise for papers where appropriate reviewers were not initially available.
The reviewing process accepted 643 papers (29.9% of valid submissions). 83 of these were accepted as oral presentations (3.9% of valid submissions) and 123 were accepted as spotlight presentations (for a total of 9.7% of valid submissions with live presentations). All papers will also appear in the interactive poster sessions. CVPR 2016 introduces a new form of presentation -- "Spotlights". Spotlights are short oral presentations aimed at highlighting the main contributions, novelties and results of papers. Our rationale behind adding spotlight presentations to CVPR 2016 was to increase the number of works that get presented to a large audience thus gaining maximum exposure and visibility. We believe that this new presentation format (oral/spotlight/poster) will make CVPR 2016 a more exciting conference.
The review process was similar to previous years. The Program Chairs made an effort to assign papers on similar topics to the same set of Area Chairs (ACs), so that related papers could be compared directly. To achieve this, between 70 and 130 papers were pre-assigned to each AC with the help of the Toronto Paper Matching System while accounting for author provided suggestions. ACs then went over the paper titles and abstracts and bid on those they would like to handle. Using all of this information as input, ACs received a final assignment of around 30 papers for which they provided ranked lists of candidate reviewers. ACs were asked to suggest candidate reviewers who were experts in the topic, had a broad view of the field, were reliable, and could provide high quality reviews. Their choices were additionally guided by the reviewers' bids (the process followed for reviewer bidding was similar to that used for ACs).
For 99.9% of the papers all 3 of the assigned reviewers were amongst those listed by ACs as candidate reviewers. For 70% of the papers at least 2 of the assigned reviewers were ranked by ACs amongst their top 3 candidates for that paper, and for 95% of the papers at least 1 reviewer was ranked amongst the top 3 candidates. Each paper was reviewed by at least three reviewers and considered by at least three ACs before a decision was made. Borderline papers and candidate orals and spotlights were discussed in groups of three non-conflicted ACs with common areas of expertise. Oral and spotlight recommendations were made by panels of 12 ACs after extensive discussion.
The Program Chairs did not submit any papers to CVPR 2016, allowing them to avoid direct conflicts throughout the review process. This year, General Chairs were allowed to submit papers, and therefore did not have any software access to the CMT system beyond that of an author. Additionally, CVPR 2016 is the first CVPR to use Researcher.cc to identify conflicts between reviewers and authors. Reviewers and ACs were excluded from any access to papers from their research groups, affiliated institutions, or collaborators. The double-blind nature of the CVPR review process was thus strictly maintained throughout.
Since its inception in the 1960s, Computer Vision has grown into one of the most important areas of research and development in modern artificial intelligence and computer science. As we observe the blossom of computer vision technologies applied in industry, we take this opportunity at CVPR 2016 to celebrate the coming of age of computer vision, and to facilitate an even closer tie between our academic world and the industry world. As a result, for the first time in CVPR's history, we are providing an exciting, "trade-show" like atmosphere to foster maximal visibility and exposure for each on-site exhibitor from promising startups and creative standouts to the biggest industry leaders. The technologies developed by these companies range from software to hardware, and demonstrate the great impact of computer vision on different industry areas. As of now, more than 70 companies have signed up to participate in our Expo. In addition to on site booths, we also offer a unique web-presence video this year, enabling recruiting opportunities and technology highlights. So far, over 20 companies are committed to participate in this special format of presence. Furthermore, companies in the computer vision industry are showing strong interest in sponsoring the CVPR community. Together they have contributed a historical amount of sponsorship funding. Additionally, more than 10 companies showed interest in sponsoring promotional items, resulting in a significant amount of additional fundings for the conference.
CVPR 2016 has also made history is its diverse organizing committee. A record number of women are involved, from General Chairs to Area Chairs. The field of computer vision has historically been male-dominated. We are pleased to see a growing number of women and under-represented minority researchers in our community. CVPR 2016 has made an explicit statement to celebrate their contributions to our field. We hope that at the conference, you get to know some of the women leaders of our field, and encourage more women and under-represented minority students to study and work in this exciting field. We truly hope that this is only the beginning!
Last but not least, we wish to thank all members of the Organizing Committee, the Area Chairs, reviewers, emergency reviewers, authors, and the CMT and TPMS teams for the immense amount of hard work and professionalism that has gone into making CVPR 2016 a first rate conference. Our thanks also go to the organizers of previous CVPRs, in particular Rene Vidal and Andrew Zisserman, for their helpful advice and support, and to Ari Kobren for a huge effort maintaining Researcher.cc. A few people working behind the scenes deserve special thanks. Yipin Zhou and Arsalan Mousavian provided critical technical assistance with CMT and managing the largest CVPR ever. It could not have been a success without them. Nicole Finn and Liz Ryan provided critical support to the general conference organization and financial well being of CVPR. Tali Treibitz and Dana Berman have provided high quality and rapid turnaround of the CVPR web pages. Eric Mortensen and Kate Saenko have again lent expertise developed over many years to shepherd the papers through the publication process. A special thank-you should also go to the Corporate Relations team - Jia Li, Kristin Dana, Brendan Morris and Andrea Frome. They have worked tirelessly to put together the inaugural CVPR Expo and to solicit industry support. Greg Mori worked with a number of teams and contributed tremendously to local arrangements, the AC meeting, CVPR website, and industry Expo. Last but not the least, this CVPR would not be possible without the record amount of support from our corporate sponsors and participants. Finally, we wish all the attendees a highly stimulating, informative, and enjoyable conference.
Program Chairs: Lourdes Agapito, Tamara Berg, Jana Kosecka, Lihi Zelnik-Manor
General Chairs: Tinne Tuytelaars, Fei-Fei Li, Ruzena Bajcsy