Spatial AI,
Turn existing camera systems into real-time spatial intelligence.
Add depth, 3D, and spatial information to existing visual systems — without the cloud.
What can Spatial AI do with existing camera systems?
See depth from a single camera
Understand spatial relationships
Convert 2D video into real-time 3D
Improve visual recognition for robots single camera
Add 3D information to medical imaging
Upgrade existing camera-based systems
1.Use Edge AI
Generate depth information in real time using external Edge AI hardware processing, without cloud connectivity or cloud-based computation.
2.Existing Cameras & Equipment
Leverage existing cameras and imaging equipment without requiring system replacement. Generate 3D video from a single camera feed, with support for configurations using two or more cameras.
3. Real-Time Performance — 30 FPS
Support real-time 2D-to-3D conversion at up to 30 FPS, targeting a smooth display experience with minimal processing delay from video input to 3D output.
4. Glasses-Free 3D Displays
Compatible with glasses-free 3D displays, including eye-tracking-based solutions designed for both single-viewer and multi-viewer experiences across multiple displays.
5.Real-Time 3D Conversion with Audio
Flexible technology integration through SDKs, APIs, and licensing options for medical devices, imaging systems, display manufacturers, and software developers.
6.Real-Time 2D / 3D Object Differentiation
Distinguish selected objects from the surrounding scene and display them in 2D or 3D within the same 3D display environment, enabling flexible visualization of different objects in real time.
7. Automatic Occlusion Correction
Automatically correct occlusion artifacts and improve depth consistency when objects overlap or scene conditions change, including challenging situations such as sudden bleeding or gauze appearing during medical procedures.
8. SDK / API / Licensing
Flexible technology integration through SDKs, APIs, and licensing options for medical devices, imaging systems, display manufacturers, and software developers.
Medical and Healthcare
Endoscopy and Surgical Procedures: Converts real-time images of the digestive tract and body cavities into 3D, helping surgeons assess the depth of tissues and surgical instruments
This technology adds depth and spatial information to medical images to assist healthcare professionals in interpretation and procedures. Using AI, the system estimates depth, the front-to-back relationships of objects, and three-dimensional spatial relationships in real time from 2D images such as endoscopic, fluoroscopic, angiographic, and fundus images. We are also developing applications for 3D visualization, image analysis, and surgeon assistance systems.
- Endoscopy: Visualizes the depth and spatial relationships of mucosal surfaces, organs, and lesions in real time to aid in observation and treatment.
- X-ray fluoroscopy: Estimates the depth and spatial relationships of organs, catheters, and other structures to assist with interventional radiology (IVR) and device manipulation
- Angiography: Provides a three-dimensional view of the course, branching, and depth of blood vessels to assist with catheter manipulation and endovascular procedures
- Fundus Images: Estimate the three-dimensional structure and depth information of the retina, blood vessels, and lesions to support ophthalmic examinations and treatment
Industrial and Manufacturing Inspections
Enhancing Recognition Capabilities on the Manufacturing Floor by Incorporating Depth Information into Image Inspection
Based on camera footage from production lines and inspection equipment, the system estimates the position, shape, depth, and relative positions of parts. It can be applied to advance manufacturing processes, such as visual inspection, part positioning, anomaly detection, and robotic picking.
1. Visual and Quality Inspection Detection and evaluation of scratches, defects, deformation, foreign objects, and surface conditions on products and parts using AI
2. Dimensions, Shape, and 3D Inspection: Measure and verify the depth, height, shape, positional deviations, and assembly status of parts based on spatial data.
3. Recognizes the position, orientation, and relative alignment of parts on production lines and during robotic inspection in real time to support picking, assembly, transport, and robot control
In points 2 and 3 above, adding depth and spatial information to existing 2D camera footage has the potential to enhance conventional image recognition
Semiconductors (Back-end Processes)
Enhancing Recognition Capabilities on the Manufacturing Floor by Incorporating Depth Information into Image Inspection
Based on camera footage from production lines and inspection equipment, the system estimates the position, shape, depth, and relative positions of parts. It can be applied to advance manufacturing processes, such as visual inspection, part positioning, anomaly detection, and robotic picking.
Main Areas of Testing
- Inspection of Chip Position, Tilt, and Mounting Condition
- Inspection of Wire Bonding and Joints
- Inspection of the Height, Shape, and Position of Bumps (Connection Terminals)
- Inspection of the Height, Shape, and Position of Bumps (Connection Terminals)
Automobiles
3D Navi（Real-Time 3D Navigation）
Based on footage from in-vehicle cameras, the system estimates the depth and spatial relationships of vehicles, pedestrians, obstacles, and the road environment. It is expected to find applications in environmental perception, driver assistance, in-vehicle displays, and next-generation mobility systems.
Improved Visibility of Potential Hazards: Displays pedestrians, bicycles, and vehicles in 3D, allowing for an intuitive understanding of the surrounding environment.Understanding Distance and Position: Makes it easier to recognize the relative front-to-back positions and depth between the vehicle and surrounding moving objects.Hazard Prediction Assistance: Helps drivers notice approaching pedestrians, bicycles crossing intersections, and other hazards more easily. Organizing Information While Driving: Keep the stationary background in 2D to make moving objects stand out.
Real-Time 3D Dash Cam
3D Dashcam: Records moving pedestrians, bicycles, and vehicles in real time as 3D models. It allows you to visually confirm the depth and relative positions of objects in three dimensions, helping you understand and analyze accident scenarios.
Three-Dimensional Recording of Hazardous Situations: Recording the movements of pedestrians, bicycles, other vehicles, and more as 3D video Understanding the Accident Situation: It is easy to visually understand the relative positions and sequences of vehicles, pedestrians, and bicycles.Visualization of Approaching and Interrupting Objects: Makes it easier to understand the movements and relative positions of nearby moving objects Advanced Accident Analysis: In addition to standard 2D footage, it enables three-dimensional assessment of the situation using depth information.
Robotics
AI understands spatial relationships in the real world and translates them into robot movements
Based on images from a monocular camera, the system estimates the position and depth of objects, as well as their relative positions relative to the robot, in real time. It is utilized for tasks such as picking, transport, positioning, and verification of work results performed by the robot arm, with the goal of realizing an autonomous system that repeatedly performs recognition, decision-making, motion, and verification. .
In robotics, there are methods for acquiring spatial information using stereo cameras, LiDAR, RGB-D cameras, and other devices. Our company, on the other hand, is able to estimate depth from existing monocular RGB camera footage.NVIDIA provides the computing infrastructure that powers AI, while our company provides a real-time AI engine that enables robots to understand their surroundings.It is possible to integrate Novias' technology into the NVIDIA ecosystem.
Monocular depth estimation is an AI technology that estimates depth and spatial information from a single 2D camera image. Unlike conventional stereo vision, it does not require two cameras or a dedicated depth sensor. NOVIUS applies monocular depth estimation to generate real-time depth information from existing camera systems.
NOVIUS uses AI-based depth estimation to generate depth information from conventional 2D camera input in real time. The depth information is then used to create a 3D representation of the captured scene. This enables real-time 2D-to-3D conversion without requiring a stereo camera, LiDAR, or other dedicated depth sensor.
Yes. NOVIUS can generate depth information from a single camera feed using AI-based monocular depth estimation. An external edge AI device processes the camera input in real time, enabling 2D-to-3D conversion without replacing the existing camera system.
Yes. NOVIUS technology is designed for real-time processing using an external edge AI device, without relying on cloud processing. Video data can be processed locally, helping reduce latency and supporting applications where real-time performance and data privacy are important.
Yes. NOVIUS technology is designed to work with existing cameras and imaging equipment without requiring a complete system replacement. An external edge AI device can add real-time 3D conversion capabilities to existing camera-based systems.
NOVIUS technology can work with conventional RGB camera systems, depending on the application and technical requirements. It can support single-camera and multi-camera configurations and is designed to add depth and spatial information to existing imaging systems.
Yes. NOVIUS can provide its technology through SDK, API, and licensing models. Our platform is designed to support integration across multiple devices and industries, allowing partners to incorporate real-time depth and spatial information capabilities into their own products and applications.
Yes. NOVIUS can work with companies to evaluate the technology through a proof of concept (PoC) or technical validation. We can discuss the target application, existing equipment, technical requirements, and evaluation objectives before defining an appropriate validation approach.
NOVIUS technology can be applied across multiple industries, including medical and healthcare, robotics, industrial and manufacturing inspection, semiconductors, automotive, and imaging systems. The platform is designed to add real-time depth and spatial information to existing camera-based systems.
Let’s Explore What’s Possible.
Discuss your challenge with our technology team and explore a PoC, technical validation, or collaboration.