l Uses images as data sources, supporting image analysis for various targets like humans, cars, and non-motor vehicles.
l Supports 1:1 comparison and 1:N retrieval, ideal for high-precision scenarios like facial, iris, and palm vein recognition.
l Enables distributed deployment and dynamic scalability, ready for instant use upon power-up.
l Provides Rest Full interface, which is efficient and easy to use.
l Supports GA/T1400 standard protocol, seamlessly connects with upstream and downstream application platforms.
Supports ARM and X86 processor architectures, compatible with various Al chips like Ascend and NVIDIA.
l Using video streams as data sources, supports structured video analysis for people, cars, non-motor vehicles, and more.
l Enables real-time Al visualization, ideal for live target tracking and cross-camera tracing.
l Manages Al algorithms through plug-ins, enabling customization video analysis tasks for cost-effective solutions.
l Supports standard video protocols like RTSP, ONVIF, for sameless integration with third-party application platforms.
Supports ARM and X86 processor architectures, compatible with various Al chips like Ascend and NVIDIA.
l Scanning Resolution: 2048x(32~50000)
l Movement Speed: 0-0.8 m/s continuous adjustable
l Turning Radius: zero radius, turn in place
l Endurance: >4 hours
Reference Weight: 12kg
l Applicable Width: 600-1200mm
l Intelligent Self-check: Automatic identification and alarm of abnormal state.
l Emergency Function: Support manual quick opening for power failure.
Open Interface: Support SDK secondary development.
l Video Anti-tailgating
Effectively detect people and objects.and solve the problem of squatting.crawling, following and other tailgating.
l Bi-directional Pass
Support for custom pass mode, can be setup one-way or two-way pass mode.
l Liveness Detection
Support live face and fingerprint detection.
l Traffic Interlock
Double doors interlock, the first door open,the second door closed,single person inspection.
l Alarm Function
Illegal entry,tailgating alarm, reverse passing,passing timeout,etc.
l Open Interface
Support SDK interface for secondary development.
l Channel Design
Ultra-thin body, single machine free combination,channel width 900mm (customization).
l Power-on Self-check
Power-on self-check aromatically.
l Alarm Function
Illegal entry / reverse passing / tailgating / passing time-out etc.
l Automatic Reset
lf the person does not pass in the specified time, it will automatically close
l Automatic Anti-pinch
Infrared/force touch dual anti-trap function.
l Open Interface
Support SDK secondary development.
l Channel Design
Ultra-thin body, single machine free combination,channel width 600-900mm
l Power-on Self-check
Power-on self-check automatically.
l Alarm Function
Illegal entry / reverse passing / passing time-out etc.
l Automatic Reset
lf the person does not pass in the specified time, it will automatically close.
l Automatic Anti-pinch
Infrared/force torch dual anti-trap function.
l Open Interface
Support SDK secondary development.
l Feature Detection: Supports mask-wearing detection, face covering reminding (voice, text reminding).
l High True Accept Rate and Low False Match Rate: FAR (False Acceptance Rate)<0.01%,FRR (False Rejection Rate) <1%.
l Liveness Detection: With live face detection function, the use of cell phone photos, paper photos and other non-real face comparison can not be compared to pass, and issue an alarm prompts.
l Environmental Adaptability: Automatic adjustment to the circumstances (e.g., intense light, backlight, weak light)
l Breakpoint Transmission: Data is stored locally after a network outage and automatically renewed when the network is up and running.
Remote Management: Remote monitoring view of the devices overview, all the operational parameter settings, and it’s detailed information; remote operation and maintenance are allowed: the face database can be updated remotely.
l Authentication Method: Support combination of human face, password, card.
l Maximum Authentication Capacity: More than 30,000 human faces and 10,000 cards; more than 100,000clearance records can be stored.
l Face Recognition: Independently developed face recognition algorithm empowered by deep learning; comparison time <0.2s in 1:N scenario with accurate rate >99%
l Feature Detection: Supports mask-wearing detection, face-obscuring alerts. (voice, text alerts)
Liveness Detection: With live face detection function, the use of cell phone photos, paper photos and other non-real face comparison can not be compared to pass and issue an alarm prompts
l Authentication Method: Support combination of human face, password, card.
l Maximum Authentication Capacity: More than 30,000 human faces and 10,000 cards; more than 100,000clearance records can be stored.
l Face Recognition: Independently developed face recognition algorithm empowered by deep learning; comparison time <0.2s in 1:N scenario with accurate rate >99%
l Face-masking Attribute Detection Detection: Support face covering alerts such as wearing masks, sunglasses, hats.etc. (voice and text alerts)
l Liveness Detection: With live face detection function, the use of cell phone photos, paper photos and other non-real face comparison can not be compared to pass and issue an alarm prompts
l Palmprint/Palm Vein Multimodal Recognition
l User Capacity 100 People
l Static Power Consumption <1uA
l Peak Power Consumption <2.0W
l Time from Cold Start to Successful Recognition <1s
l Standard TTL Serial Communication for Easy Integration
Industry-universal PCB Size and Screw Hole Design
l Non-contact Capture
l Venous Live Imaging
l High Capture Quality, Anti-interference
l Binocular 200W pixels
l USB 2.0 High-speed Interface
Support Secondary Development, including Windows, Linux, Android, etc
l Non-contact Capture
l Venous Live Imaging
l High Capture Quality, Anti-interference
l USB 2.0 High-speed Interface
Support Secondary Development, including Windows, Linux, Android, etc.
l Palm Vein/Palmprint Recognition
l Recognition Speed <0.3s
l Wet Hand can be ldentified
l Anti-Motion Blur
l Palm Offset can be ldentified
l Water Mist Environment can be ldentified
l Recognition Speed <0.3s
l Liveness Detection
l Wet Hand Recognition
l Protection Level IP54
l High Recognition Rate
l Non-contact Vein Recognition
l Different Angles of the Palm is Recognizable
Recognizable Under Water Mist Environment
On January 22, 2024, Maxvision once again achieved outstanding results in the Face Recognition Technology Evaluation (FRTE) competition, the most authoritative face recognition algorithm competition announced by the National Institute of Standards and Technology (NIST) in the United States. In key metric tests, Maxvision's algorithm secured the 5th position in Mugshot-Profile 90° (R=1, T=0), 7th position in Mugshot-Profile 90° (T>0), 8th position in Mugshot-Mugshot AT>12YRS, and 10th position in Visa-Border. Maxvision's face recognition algorithm excelled in all four key metrics, placing it among the top 10 internationally.
FRTE is renowned as the world's largest, most rigorous, highly competitive, and authoritative face recognition algorithm competition. It is considered the gold standard in the industry, with evaluations reaching an accuracy of one in a million. To date, over a hundred companies and research institutions worldwide have participated in this competition, including VisionLabs (Russia), Ever AI (USA), Kakao (South Korea), NEC (Japan), and Idemia (France).In contrast to some companies focusing solely on a single scenario, Maxvision's face recognition algorithm demonstrated balanced development across multiple scenarios in this competition, performing exceptionally well in various scenarios, showcasing strong generalization and wide applicability.
In the various track tests participated by Maxvision Technology, Mugshot-Profile 90° represents the facial comparison at a 90-degree (extreme angle) view, demonstrating the algorithm's performance across different facial angles. In situations where the camera can only capture side-face 90° information, compared to the previously collected and stored frontal facial information, there is a higher degree of feature loss, making comparisons more challenging.Mugshot-Mugshot AT≥12YRS represents the comparison with an age difference of over 12 years, illustrating the algorithm's capability for facial recognition across different ages. As age increases, facial appearance features undergo changes such as wrinkles, spots, and skin sagging, which can make accurate recognition difficult, especially when dealing with age differences exceeding 12 years.
Visa-Border represents the comparison of visa photos with entry and exit photos, showcasing the algorithm's performance in scenarios like ports and airports. The primary challenge lies in handling the comparison tasks between visa photos and on-site inspection photos. Factors such as lighting conditions, changes in perspective and pose, and facial similarity among relatives make accurate identification highly challenging. Maxvision Technology's exceptional performance in these track tests demonstrates its strength and innovation in the field of facial recognition technology.
As a company specializing in artificial intelligence and machine vision technology, Maxvision Technology has been dedicated to innovating and developing biometric recognition technologies such as facial, iris, and palm vein recognition, continuously elevating its technological capabilities. Maxvision's AIGC image generation enables the creation of content with few or no samples, significantly enriching the diversity of produced content. Simultaneously, Maxvision's multimodal AI recognizes, transforms, and generates different modal data (text, images, audio, etc.), leading a new AI interaction paradigm and enhancing the universality and creativity of multimodal fusion AIGC.Maxvision's large-scale models in the smart port industry strive to create exclusive large models that not only possess general knowledge and capabilities but also deeply integrate domain knowledge specific to the port industry and scenes. These models are gradually being applied in a wide range of business scenarios, including duty command, front-end inspection, business consultation, image recognition, and risk analysis.
The outstanding achievements Maxvision Technology attained in this international competition not only validate its technological prowess but also showcase its unwavering commitment to innovation and excellence Maxvision Technology will continue to leverage its advantages in artificial intelligence technology, continually improving the accuracy and reliability of biometric recognition technology. This empowerment aims to serve more intelligent scenarios and, through Al-driven industry innovations, sustainably lead and propel industry development.
Maxvision Technology Corp. (stock code "002990") is a Shenzhen Stock Exchange-listed enterprise specializing in artificial intelligence, big data, and IoT. As a national high-tech company, it offers intelligent products and "AI+Industry" solutions. With 27 years in the port field, Maxvision has over 50% R&D staff, an independent center, and 1,500 intellectual property rights.
Plan your visit to the trade fair!
