Remark AI's Research and Development Team Led by Dr. Xiaoyun Yang Wins Three Champion Awards at ECCV 2020
Remark AI won the championship in the Long-term tracking challenge (VOT-LT2020), the Short-term real-time tracking challenge (VOT-RT2020) and the Color and Depth long-term tracking challenge (VOT-RGBD2020). "We are proud of what we accomplished at ECCV as it reinforces that our AI products and services are based on some of the highest-quality AI software available," said
The biennial ECCV is the European continent's top event in the image-analysis field. The 2020 edition, the 16th, took place online due to the COVID-19 pandemic. The Visual Object Tracking (VOT) challenges represented the eighth annual tracker benchmarking activity organized by the VOT initiative and held in conjunction with the ECCV. Co-founded by the
The VOT challenges evaluate the performance of each participant's algorithm for short-term tracking of single targets in complex scenarios. Evaluations include the standard VOT and other popular methodologies for short-term tracking analysis and a "real-time" experiment simulating a situation where a tracker processes images as if provided by a continuously running sensor. The long-term tracking sub-challenge focuses on long-term tracking properties, namely coping with target disappearance and reappearance. Since the evaluation sequence is updated every year, and the accuracy of the annotation is improved year by year, the VOT competition is also regarded as the most difficult competition in the field of visual tracking, far surpassing other data sets.
In recent years, the competition has become extremely competitive, with many of the largest technology companies, including Microsoft Research Asia and
Advancing AI technology Research & Development for wider applications
Remark AI is an innovative enterprise dedicated to the implementation of advanced artificial intelligence technology. The company has established laboratories jointly with well-known foreign universities such as
Remark AI continues to break new ground in the research and development of cutting-edge technologies. In recent years, it has successively won leading rankings in multiple international competitions in the field of video tracking and segmentation, marking Remark AI's industry leading position in key video intelligent analysis technologies such as video target detection, tracking and segmentation technologies. At the same time, these technologies will be widely used in the 5G+AR video business of Remark AI's smart retail, smart campus, smart construction, smart city, internet of vehicles, medical and other fields.
Getting the world back to work sooner and safer
During the ongoing COVID-19 pandemic, Remark's AI thermal imaging systems provide solutions to identify individuals with higher than normal skin temperatures, and to protect other individuals from possible infections. According to CDC, fever is the earliest symptoms of COVID-19. Remark Thermal products are industrial grade and developed to efficiently and accurately detect raised body temperatures of individuals being scanned, providing a first layer of security.
The use of thermal scanning can mitigate the risk of spreading any type of contagious illness, but also include a multitude of other benefits such as:
- Improve the safety of staff, and cut down operational costs from not having a trained staff member to individually scan each person entering with an outdated temperature gun;
- Have the ability to view alerts in real time both on display and can also be sent as an email or text to authorized administrators;
- Create a discreet and non-intrusive experience for your customers and staff;
- Integrate with employee check-ins and contactless gate and door access;
- Provide contact tracing questions after a temperature scan.
In addition to video tracking and thermal imaging, Remark AI has also made new progress in facial analysis. In response to the current situation where people generally wear masks, mask wearing detection, face recognition under masks, and attribute analysis models have been developed. In addition, infrared thermal imaging technology has been integrated to form an independent "Remark Brain" integrated platform to help society solve all kinds of new face problems under the new epidemic prevention situation.
With the resumption of work and schools, epidemic prevention for people in public places such as office buildings, hotels, hospitals and schools become even more challenging. The movement of people has brought economic activity and greater health risks. Remark AI's series of mask-wearing face models provide complete solutions for all sectors of society in a timely manner. Its functions include mask detection for employees, non-contact temperature measurement, and daily face-scanning access control for check-in and check-out.
In addition to individual settings, according to a test done on 100,000 actual pictures, the model can quickly identify and label faces wearing masks and those without masks while detecting mass faces in densely crowded public scenes. The face detection accuracy rate reaches 98.6%, and the mask judgment accuracy rate reaches 96.8%. The test is done within 10ms, the industry's leading level. Behind such a high accuracy rate is the ingenious model design and the result of training with a large amount of data. On one hand, the new model uses multiple data sets, training data of more than 100,000 images, and reconfirms clearly to ensure that the sample size is large enough and the annotations are correct and effective. On the other hand, the face detection model is based on a new algorithm independently developed by Remark AI. The entire R&D process is based on the "Remark Brain" deep learning platform, which enables efficient and convenient model development, training, and deployment. For the lighting, mask obstruction, expressional changes, angle changes, scale changes and other issues in the actual implementation, the model is robust and can be detected in real time on a variety of different end, edge, and cloud devices.
This press release may contain forward-looking statements, including information relating to future events, future financial performance, strategies, expectations, competitive environment and regulation. Words such as "may," "should," "could," "would," "predicts," "potential," "continue," "expects," "anticipates," "future," "intends," "plans," "believes," "estimates," and similar expressions, as well as statements in future tense, identify forward-looking statements. These statements involve known and unknown risks, uncertainties and other factors, including those discussed in Part I, Item 1A. Risk Factors in
Director of Capital Markets and Investor Relations
Vice President of Investor Relations
View original content to download multimedia:http://www.prnewswire.com/news-releases/remark-ais-research-and-development-team-led-by-dr-xiaoyun-yang-wins-three-champion-awards-at-eccv-2020-301121798.html