Recently, the ICDAR 2023 events have come to an end, and the Document Image Analysis and Recognition team from CVTE Central Research Institute has won multiple championships and runner-ups!

A number of remarkable outcomes in top-level international competitions

ICDAR (International Conference on Document Analysis and Recognition) is one of the most important international conferences in the field of document image analysis. ICDAR has organized more than ten competitions this year, drawing participation from leading technology companies and prestigious colleges from around the globe, including Google, Amazon, Baidu, Alibaba, Tencent, Peking University, and Tsinghua University. In this competition, the CVTE team won first place in the end-to-end video text recognition task of BDVT-QA (V-DA) and second place in all three tasks of the CROHME handwritten formula recognition event - On-line recognition, Off-line recognition, and Bimodal recognition (YP_OCR).

Competition tasks involving text recognition and formula recognition are closely related to the business technologies developed by CVTE for the education sector. These technologies are now extensively used in Seewo learning machines for homework correction and mental arithmetic calculation products. The product performance and user experience have both been significantly enhanced thanks to the maturation and optimization of underlying technologies, providing a strong base for the reputation of products.

Flourish in a variety of sectors, exceptional technical strength

The Central Research Institute was initially founded to provide strong support for the company's technology growth strategy. Currently, 25% of the researchers at the Central Research Institute hold doctoral degrees, with their primary research interests being in the fields of visual computing, speech signal information processing, tactile technology, spatial perception, natural language processing, medical signal processing, and data mining. The team now consists of top-tier international talents from universities such as UCLA, Tsinghua, the Chinese Academy of Sciences, Apple, and others. In their spare time beginning in 2018, the various divisions of the Central Research Institute started competing in technical competitions that were closely related to their business products and were successful in taking home many honors with their strong capabilities:

·China Conference on Knowledge Graph and Semantic Computing Evaluation Task: Command Understanding Task over the Music Domain (3rd place, 2018)

·Alibaba Cloud Tianchi: "Digital and Smart Education" Data Visualization Innovation Competition (1st place, 2019)

·ACM MM Challenge AI Meets Beauty (3rd place, 2019)

·Alibaba Cloud Tianchi: 2nd Hainan Big Data Innovation Application Competition - Intelligent Algorithm - Resume Parsing Contest (5th place, 2020)

·2021 iFlytek AI Developer Competition - Question Label Prediction Challenge (2nd place, 2021)

·CCL 2022 Chinese Learner Text Correction Competition (3rd place in Track 1, 2nd place in Track 4, 2022)…

In addition, the Central Research Institute will also reserve and verify future technologies through competitions. This year, the data mining team (CVTEDMer) from the Central Research Institute took part in the Huawei Causal Inference Challenge (PCIC). They won online first place and finished second overall in the final defense after the online screening and final defense. The models participating in this competition can plan for the future based on current business, predict failures according to product usage, and prepare solutions in advance.

Consider the big picture and make active preparations for multimodal interaction and perception

The research directions of the Central Research Institute cover a wide range of fields. Looking at the development trend of technology and combining the team's advantageous research areas, we are currently concentrating on research and development and actively deploying multimodal perception and interaction technologies. With the rise of internet technology and e-commerce trends, AR technology is becoming increasingly popular in virtual fitting applications. In response to the pain point of not being able to experience clothing materials in online shopping and consumer expectations for future online shopping, the Central Research Institute has made early arrangements and delved deeply into texture feedback technology, extending and exploring more application scenarios for texture feedback.

In recent years, gesture control has been increasingly featured in various applications due to its natural, efficient, and convenient advantages. Consumers particularly appreciate the safety of contactless operation and not directly touching objects. The Central Research Institute relies on its research advantages to conduct in-depth research and development on gesture interaction, striving to apply it to more scenarios, fields, and devices, and to refresh people's multi-modal interaction experience. Vision is a crucial aspect of human perception since it enables us to comprehend our surroundings, identify gestures, recognize different face expressions, track eye movements, and more. With "See the world clearly and understand the world" as its vision, the vision research at CVTE's Central Research Institute focuses on developing a more diverse field of vision research and application by conducting in-depth research in areas like medical imaging, 3D scene perception, virtual humans, and emotional intelligence.

In the field of voice interaction, we conducted in-depth research on the audio pickup module in combination with business scenarios, significantly improving the audio pickup quality in classroom and conference settings. Through the research on voice recognition and semantic understanding, we have realized the platformization of the technical solution, which can provide more convenient and intelligent voice interaction for various smart terminal devices.

Forge ahead, steadily incubate emerging businesses

Relying on the deep research and practical experience of the Central Research Institute in basic and applied technology, strategic incubation is currently underway in several emerging fields. We hope that the research achievements and innovative thinking of the Central Research Institute will incubate more emerging businesses in the future and cultivate a new generation of scientists and entrepreneurs. In a broader context, we look forward to more imaginative and research-oriented individuals joining us to strengthen the Central Research Institute, create more businesses through technological incubation, and turn dreams into reality.

CVTE is always actively creating a progressive, inclusive, and open research environment, closely following the trends of cutting-edge technology, vigorously promoting the transformation of research results in future education, enterprise services, intelligent hardware, health care, and other fields. We look forward to fully realizing the industrial and social value of technology. CVTE will also continue to uphold the belief in empowering through technological innovation, making our presence increasingly vital in helping more people achieve success in their careers and happiness in their lives.