AddaxAI: A no-code platform to train and deploy custom YOLOv5 object detection models
Peter van Lunteren
Abstract
EcoAssist is a tkinter based graphical user interface (GUI) which is designed to enable object detection for novices.The package allows the user to annotate images, train and deploy custom models, and post-process imagery without having to write code (Figure 1).Installation is automated and dependencies will be installed in a virtual environment to avoid conflicts.Annotation is done via the open-source package labelImg (Heartexlabs, 2022) and postprocessing features include folder separation, detection visualization, cropping, label creation, and exporting results to CSV files.It will automatically run on NVIDIA or Apple Silicon GPU and is available for Microsoft Windows, Apple macOS, and Linux.The main target audience is ecologists working with motion triggered camera traps, although it is not limited to detect animals as object detection models are generic.The package is developed to be used in conjunction with Timelapse (Greenberg et al., 2019) (an image analyzer for camera traps) and MegaDetector (Beery et al., 2019) (a trained object detection model to localize animals in camera trap imagery).More features are planned and will be added in the future.