Some applications can greatly benefit from characterizing the size distribution of droplets suspended in an emulsion. Coanda has developed a custom measurement technique that can automatically detect these droplets.
A specialized off-the-shelf microscopy unit takes images of the emulsion. The images are then processed using a convolutional neural network (ConvNet) to identify the size and location of the droplets. The ConvNet has been pretrained on public datasets and fine-tuned on about 100 of the application images. It can find the droplets in under a second.
Sometimes, the droplets can be larger than the microscopy unit’s field of view. In these cases, the observed interfaces are located using a ridge-finding algorithm. For the water-in-oil emulsion pictured here, the ConvNet identified droplets are shown as red boxes, and the interfaces as green lines. Together, the ConvNet and ridge-finder create a hybrid processing technique that is used to calculate important statistics about the application.
This custom algorithm is based on state-of-the-art techniques for image classification and object detection.