converts video into structured data
Automated Video Analysis Software
Our v3 Tools software is powered by proprietary parallax induction and segmentation image analysis algorithms. v3 Tools is a mathematical engine that directly converts video into structured labeled data.
v3 Tools automatically detects, tracks, locates, characterizes, and classifies scene objects. It extracts detailed information about each object, including its location, velocity and heading, size, shape, color, and track. The information includes track vectors and shape files for all the objects.
Suitable for Machine Learning Validation and Training
Segmented individual objects shape files are stored on a frame by frame basis in a manner suitable for Machine Learning validation and training.
This intelligence data is extracted for each object in each frame of video, and is written to a single relational database file. The relational database allows objects attributes to be analyzed and compared within a frame, or across the entire Enterprise.
v3 Tools also locates object shape files directly onto the true target locations with excellent correspondence (providing object location verification), for all three FMV modes: EO-NFOV, EO-WFOV, and IR.
Hundreds of billions of dollars are currently being spent to develop AI and ML applications. While significant progress has been made in making AI and ML applications a reality, a machine with the ability to fully comprehend live video streams in real time is still lacking. Most current AI and ML applications require a structured labeled image database to function. The majority of structured image databases are currently labeled manually by humans. This requirement is one of the fundamental limitations slowing AI adoption down.
The Company has developed the v3 Tools video and image analysis software that automatically generates a structured labeled image database in real time from video and image sources. v3 Tools is in part based on the Company’s PV and IS technologies.
v3 Tools is capable of automatically generating structured labeled data from video sensor streams. v3 Tools does not require a database or excessive computing power in order to process video. The v3 Tools database is compatible with current AI industry data standards.
The ability to automatically generate structured labeled data from streaming video is a fundamental necessity for any real time autonomous AI application.
v3 Tools Markets include:
Intelligence Surveillance Reconnaissance (ISR) Sensor Data Analysis
Facility and Perimeter Surveillance
AI & ML Image Chip & RGB Training Data Generation
v3 Tools is a Mathematical Engine
Fully automated analysis of Full Motion Video (FMV) and other sensor imagery data
Includes streaming video, thermal IR, WAMI and satellite imagery
Automatically detects and tracks objects in each frame of video, counts, geo-locates, and measures each object’s properties, and writes the data from each frame to a database
Database information can be output and visualized in specific modes (tracking, shape, segmented and image chips)
Automatically generates large training datasets in many formats, including bounding boxes, semantic segmentation shape files or polygons, and pixel masks suitable for use by AI/ML applications
Detailed object properties (multi-dimensional feature space) are provided for Third Wave AI processing
In addition to its image processing capabilities, Vision III Imaging has also developed parallax visualization (PV) technologies for capturing and displaying three-dimensional information on standard unaided displays. The PV technologies create dimensional advertising that appears distinctly different from the surrounding normal video content.
Both of the Company’s IS and PV technologies are unique and have tremendous revenue generating potential in both the military and the ever-expanding commercial markets that includes such applications as autonomous (self-driving) vehicles, robotics, and medical imaging.
v3 Trafficview ™
v3 Trafficview is a simpler version of v3 Tools designed for traffic management using data generated from streaming video cameras. It counts and classifies vehicles and railroad cars in real time.