Table 2. A summary of the existing tools for automated visual tracking of animals based on qualitative features:

Installation, interface, environment, detection method, tracking method, dataset generation, animals tested on and any additional features. We compare MOTHe with a variety of different tools such as TRex (Walter & Couzin, 2021), AIDE (Kellenberger, Tuia & Morris, 2020), SORT (Bewley et al., 2016) and Koger 2023 (Koger et al., 2023).

	TRex	SORT	AIDE	Koger et al.	MOTHe
Installation mode	Command-based	NA	Web-based	NA	Command-based
Integrated pipeline?	Yes	No	No	No	Yes
GUI	Yes	No	Annotation tool	No	Yes
Supported OS	Windows, Linux, Mac	NA	Web-based	NA	Windows, Linux, Mac
Image acquisition	Video input using TGrabs	Automated	Camera trap dataset	Model-assisted labeling	Point and Click
Detection method	Background Subtraction and Neural Networks	FrCNN	Deep learning	Detectron2 API within the PyTorch framework	Grayscale Thresholding, Deep Learning (using CNNs)
Tracking method	Kalman Filter and custom tree-based method for ID	Kalman Filter and Hungarian algorithm	Not supported	Modified version of the Hungarian algorithm	Kalman and Hungarian algorithms
Animals tested	Fish and Insects	Not tested on animal videos	NA	Monkeys and African ungulates	Antelope and Wasp
Demonstration for natural conditions	No	No	NA	Yes	Yes
Max #animals	100	NA	NA	1024	156
Manual Id correction required?	No	Maybe	NA	Maybe	Yes
Extra features	Posture analysis, 2D visual fields and real-time tracking			Body postures (poses) and environmental features reconstruction