Automated object detection and image classification pipeline

A mixture of Python and R scripts were used and automated to create a data processing pipeline which runs and utilizes the output of a pre-trained object detection model (MegaDetector) to automatically classify all images in and added to the Snapshot Wisconsin project. This implementation decreased the human image classification backlog by nearly 10 million images (> 50%) and gives an initial classification to all new images within minutes of upload.

Prior to implementation of the automated classification process, it was estimated that catching up on classifying the backlog of historical images would take over 3 years with human classification options available at full capacity, not to mention keeping up on 1 million new images coming in every month. The automated classification process not only reduced the expected time to full classification of the historic data, but keeps up and filters new images to help with optimizing human workloads. More work is planned to be done in this area.