CPICS – Continuous Particle Imaging and Classification System
Our Continuous Particle Imaging and Classification System (CPICS) is an imaging sensor that produces unprecedented results for in-situ aquatic microscopy of seawater, freshwater and laboratory samples. Using darkfield illumination, the CPICS-1000-e captures high-resolution color images, showing features as small as 1µm and as large as several cm. Color information is key to high-accuracy classification while also providing important physiological information such as pigmentation due to grazing on phytoplankton. Because of its open-flow approach to water sampling, the delicate structures of plankton and particles remain completely intact as do predator-prey interactions.
The CPICS-1000-e is our latest development, which features embedded processing, region of interest (ROI) extraction, and on-board plankton classification. Stand-alone deployments on CTD Rosettes and autonomous platforms or vehicles – from large research vessels to row boats – make CPICS-1000-e an indispensable tool for researchers and resource managers alike.
CPICS-1000-e product data sheet
ROImanage – Image Data Management Software
The ROImanage software product is a browser-based application used for manually categorizing images that are collected by the CPICS family of instruments. Users can create an unlimited number of named categories into which they can drag and drop images to create data sets for further analysis or for viewing based on search criteria. ROImanage is also the base application with which our automated classification software module ROIclass operates.
ROImanage software product data sheet
ROIclass – Image Classification Software
The ROIclass software application is an add-on module that extends the manual capabilities of the ROIsort browser-based application by generating automatic classifications of images that were collected by the CPICS family of instruments. Using the manually classified Region Of Interest (ROI) sets created in the ROIsort application, ROIs are used by ROIclass to build classifiers based on Deep Convolutional Neural Networks, Support Vector Machines, and Random Forests. The user can also select from several types of machine vision descriptors that can represent the image in various ways for the training and running of the classifiers.
ROIclass software product data sheet
OceanCube – Coastal Observatory System
An OceanCube is an un-manned underwater coastal observatory system designed to provide real-time data and images. A central node supports a variety of biological, physical, and chemical sensors and is optionally connected to as many as four satellite nodes at the corners of a cubic volume that provide current and temperature information. To observe the behavior of fish, stereo camera modules with hydrophones can also be connected. A cable from a shore laboratory to the central node is used to supply power, remotely control individual instruments and to transfer data back to shore at high speed.
HabCam V-5 Vfin – Habitat Camera Vehicle
HabCam V5 is a benthic stereo-imaging vehicle built into a 4 ft. modified di-hedral v-fin frame. It is designed to be towed at a speed of 3-6 kts while taking high resolution color image pairs of
the sea floor. The HabCam V5 sensor package is customizable and compatible with most oceanographic instrumentation. The side by side stereo pair images are fused into a single image at the time of acquisition allowing precise stereo
referencing with other instrumentation’s metadata such as latitude, longitude,
temperature, salinity, chlorophyll, light absorption, dissolved
oxygen, and other environmental data. V5 also houses a side scan acoustic sonar system that collects 3D bathymetry on the range of 100 m on either side of the vehicle. These data are packetized into a gigabit Ethernet signal and transmitted to the surface, providing real-time data of the imaged environment. Image data, altitude, and depth are immediately available and visualized for a fly-by-winch pilot who operates the vehicle, adjusting the altitude as needed.