Invasive Plant Detection
We built and fielded a dual-image multispectral, visible and NIR, capture and process system to detect weeds remotely. On trigger, two images from a single camera are collected and processed to highlight the existence of skeleton weed. The camera contains two sensors that share their view through a single lens keeping the two images perfectly aligned and reducing the computational complexity. The algorithm was developed in-house by Sensory Labs.
Autism Spectrum Disorders Research
We built three synchronous six-camera systems capable of collecting imagery, as well as audio, from small classrooms to study stimulus responses in autistic children in their learning environment. The gigE cameras, with power over Ethernet capability, can be installed over 100 feet away from the recording unit. This flexibility and easy cabling gives the team many installation options. Lite compression of the video allows for longer recordings. This system also features remote frame tagging capabilities. The specialist, in the room with the child, can use a small hand-held remote IR transmitter to tag the video. The tag is used for quick recall while reviewing data to find the incident of interest. This novel non-intrusive marking system reduces additional stimulus while aiding the research.
Aerial Mapping – Extreme Environment
Sensory Labs provides high resolution cameras for an extreme environmental application. The OEM wrote the custom recording and mapping software for this application. Sensory Labs provides the cameras, electronic lens control units and consultation. The cameras are subjected to significant temperature ranges and high vibration while in use. Dozens of units are deployed.
Fish Counting
Sensory Labs was the vision software sub-contractor for a fish counting effort related to fish migration being blocked by a dam. The system uses two techniques for counting fish; a line-scan camera, and a laser sensor for use underwater. The fish are encouraged to swim into the elevator space. As they pass through a vertical light beam detection area, they are counted. Fish leaving the elevator are also detected and the count reduced. When the elevator space has sufficient fish, the chamber is closed and raised. The fish are poured onto the flume where they are de-watered. Fish pass the view of a second line-scan camera and are counted again as they enter the holding tank. Later, staff sorts the fish into a number of trucks. Some fish are taken back down the river, some are moved above the dam, and others are moved further up the river past additional barriers.
American Sign Language Database
We built a four camera system to record native sign language speakers. The cameras were combined to generate 3D video information. The data was used to create a visual dictionary for American Sign Language (ASL). The dictionary provides a reverse look-up database for translating sign information. The user signs into the computer’s camera. Software searches the extensive database. The most likely sign matches are presented. This method is useful for translating sign into text, as well as, helping people learn how to sign. If you have a word to express, it’s generally easy to find the sign. But, have you ever seen a sign and wondered what it meant? Now you can look it up!