On building a worm detector

The article discusses the process of building a worm detector to identify and track individual and collective movements of worms in a soil environment. The author uses a machine learning model to analyze data from a worm-tracking system, which involves a transparent plastic sheet with a grid of holes, a camera, and a data logger. The system can track the movement of individual worms and record their speed, distance traveled, and other metrics. The author describes the process of data preprocessing, feature extraction, and model training, including using a convolutional neural network to classify worm movements. The author also discusses the potential applications of this technology, including monitoring soil health and optimizing soil management practices.