ICCV was held in Seoul, South Korea from October 27th to November 2nd. The conference was co-chaired by Professor Kyoung Mu Lee from the Department of Electrical and Computer Engineering at Seoul National University, Professor David Forsyth from the Department of Computer Science at the University of Illinois at Urbana-Champaign, Professor Marc Pollefeys from the Computer Science Department and the Institute of Visual Computing at ETH Zurich, and Professor Xiaogang Wang, the founder of SenseTime and a professor at the Chinese University of Hong Kong. The conference officials provided information on paper submissions and attendees, as well as announced the award-winning papers of this year's conference.
At this globally significant computer vision conference, Chinese scholars shone brightly and attracted attention. Chinese researchers ranked first in the number of selected papers, and also led in terms of the number of awards won. The only regret was the absence of Chinese researchers in the Best Paper and Best Student Paper categories.
ICCV, short for IEEE International Conference on Computer Vision, is one of the top three computer vision conferences along with the Conference on Computer Vision and Pattern Recognition (CVPR) and the European Conference on Computer Vision (ECCV). It is organized by IEEE and is highly regarded in the industry, being ranked as the highest-level academic conference by the Australian ICT Academic Conference and the China Computer Federation, among other institutions.
Unlike CVPR, which is held once a year in the United States, and ECCV, which is held in Europe, ICCV is held every two years worldwide. It has a very low acceptance rate and is recognized as the highest-level conference among the top three. The conference usually lasts for four to five days, during which experts in the field will showcase the latest research findings.
ICCV 2019 was held in Seoul, South Korea from October 27th to November 2nd, lasting a full seven days. The conference has been held for 32 years since its inception in 1987.
Research in the academic community is one of the driving forces behind the rapid development of artificial intelligence technology. In recent years, artificial intelligence technology has been widely applied in fields such as security, transportation, smart cities, and education, and has gradually extended to the enterprise office field, including scenarios such as facial recognition attendance and smart conference rooms. It has also become commonplace in personal terminals, such as facial unlocking on mobile phones and voice recognition. All of this is inseparable from the research and development of basic academic knowledge. Therefore, this top conference holds great significance.
According to media reports, ICCV 2019 received 4,303 paper submissions, more than double the number of the previous conference, which had 2,143 submissions. The number of authors reached 10,000, and the total number of participants reached a record high of 7,501, a year-on-year increase of 214%.
According to statistics from Academic Headlines, among the 1,077 papers accepted at this year's ICCV, 523 were authored by Chinese researchers, accounting for 48.6%, nearly half of the total. Although the acceptance rate this year was only 25.02%, the strength of Chinese researchers is still remarkable.
At this conference, Chinese universities and technology companies performed exceptionally well. According to official statistics from ICCV 2019, China had over 350 papers selected, while the United States had over 310 papers. Germany, in third place, had less than 50 papers. Both Chinese universities and companies demonstrated outstanding performance.
According to statistics based on the first author's affiliation, the Chinese University of Hong Kong had 30 papers, the Chinese Academy of Sciences had 26 papers, and Tsinghua University had 24 papers. In contrast, top American universities such as CMU and Stanford, which have performed well in other top conferences, had fewer than 20 papers selected.
In the industry sector, China's strength is even more invincible, especially among AI unicorns and major technology companies. SenseTime and its joint laboratories had a total of 57 papers selected for ICCV 2019, while Megvii (Face++) had 11 papers selected. Although Chinese technology giants such as Huawei, Tencent, Alibaba, and Baidu did not disclose their overall data, their respective laboratories demonstrated strength. Huawei Noah's Ark Lab had 19 papers selected. Tencent's YouTu Lab had 13 papers selected. Alibaba's DAMO Academy also won a world championship at this conference, and its achievements have already been applied to Taobao's "Pai Li Tao" platform. China's strength is unparalleled.
Beijing-based SenseTime, founded in 2014 by Professor Xiaogang Wang from the Chinese University of Hong Kong, is a unicorn company in the field of artificial intelligence. It focuses on the original research and development of computer vision and deep learning technologies.
At this year's ICCV, SenseTime and its joint laboratories had a total of 57 papers selected and won 13 important competitions, including the championship in the Open Images competition, which is considered the most challenging competition in the fields of general object detection and instance segmentation. It has a much larger dataset and more complex data distribution than the COCO dataset, making it highly prestigious.
In addition to its achievements, SenseTime has also made positive contributions to the global academic community in terms of open academic exchanges and ecological development. The open-source object detection framework MMDetection, released by SenseTime and its joint laboratories last year, has been widely used by many participating teams in various competitions this year. Many award-winning teams in competitions such as Open Images, COCO, LVIS, and Mapillary have used MMDetection as a benchmark codebase. The HTC and Guided Anchoring methods proposed by the SenseTime team last year have become key drivers for many teams this year.
SenseTime's papers accepted by ICCV 2019 have achieved breakthroughs in multiple fields, including deep network basic operators for object detection, point cloud processing backbone networks based on interpolation convolution, human perception and generation for AR/VR scenes, and multimodal analysis for comprehensive scene understanding. These groundbreaking computer vision algorithms have a wide range of applications and will contribute to the overall development of the AI industry.
At the ICCV exhibition area (Booth: A-1), SenseTime showcased many innovative AI products, including autonomous driving, the SenseAR special effects engine Avatar, SenseMatrix object 3D reconstruction, and SenseMatrix facial 3D reconstruction, among others. These products provided a rich experience and innovative inspiration for the conference.
At this conference, the strength of China's presence was remarkable. As an important part of China's strength, SenseTime's demonstrated capabilities are truly impressive and deserving of applause. May SenseTime uphold its originality, ride the waves, and forge ahead.
References: Baidu Baike, Sohu, China.com, YJSV.
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.