The right kind of intelligence
Our system is built from the ground up for use in heavy industry.
Sensor
Adaptive exposure image intensity normalization to optimise detection.
Pre-processing
Automated image quality assessment and saliency to reject poor quality images and identify regions of interest.
Object recognition
Custom deep convolutional neural network (D-CNN) trained on data from real heavy industry sites for object recognition.
Object tracking
Recursive Bayesian estimation and motion modelling to track objects through time.
Object distance estimation
Stereoscopic processing for object distance estimation..
Application logic
Configurable application logic to drive desired output behaviours.
Automated health monitoring
Intelligent hardware and software self diagnostics and healing to increase system resilience.
Connection handling
Seamless edge to cloud handling to ensure operation with intermittent internet connection.
Pre-trained and pre-loaded
Our AI is pre-trained to identify common risks in industrial environments. No matter the setting, Blindsight’s computer vision knows what it’s looking for.
“Blindsight gives me video-based reporting and map locations, which we can use for internal safety training. Operators like the simplicity of the in-cab alert system to assist them around their blind spots, and the fact we can configure it to the machine and the work it’s doing.”
Truly Modern AI
Proximity sensors, pixel comparisons and beacons just don’t cut it anymore.
The core of Blindsight’s intelligence is the deep convolutional neural network (D-CNN) trained on real scenarios and tuned with machine learning for object recognition, tracking, distance recognition.
The system uses cutting edge disparity mapping for distance computation and real-time edge detection AI for alerts.