Deep learning in automated worm identification and tracking for C. elegan mating behaviour analysis

Akpu, C. H., Wei, H. ORCID: https://orcid.org/0000-0002-9664-5748 and Hong, X. ORCID: https://orcid.org/0000-0002-6832-2298 (2024) Deep learning in automated worm identification and tracking for C. elegan mating behaviour analysis. In: 27th International Conference on Pattern Recognition, ICPR 2024, 1-5 Dec 2024, Kolkata, India, pp. 113-128. doi: 10.1007/978-3-031-78122-3_8 (Antonacopoulos, A., Chaudhuri, S., Chellappa, R., Liu, C. L., Bhattacharya, S. and Pail, U. (eds.) Springer. ISBN: 9783031781216)

Abstract/Summary

This study is concerned with computer vision technologies applied in C. elegans (diminutive nematodes) mating behaviour analysis, more specifically object detection and tracking to find contacts of male and female worms in worm mating videos. Advanced deep learning algorithms, such as YOLOv8 and DeepSORT, are adapted in the automated worm identification and tracking system. A modified DeepSORT algorithm is developed to cope with appearance similarity of C. elegans for improving the tracking accuracy. In addition, a male worm detection and tracking algorithm, utilising the male worm’s mobility characteristic, assists the modified DeepSORT in accurate male worm tracking. Finally, worm contact detection is implemented by calculating the Euclidean distance between the male and female worms. The developed system, named as M1 and M2, is trained and evaluated under two sets of data, bounding boxes and segmented worms, respectively. Furthermore, we compared the effectiveness of including SAM segmentation optional module in experiments. The evaluation results have shown that YOLOv8 has excellent performance in worm detection to cope with deformable worm shape, and the modified DeepSORT significantly outperforms the default DeepSORT in worm tracking.

Item Type	Conference or Workshop Item (Paper)
URI	https://reading-clone.eprints-hosting.org/id/eprint/119635
Identification Number/DOI	10.1007/978-3-031-78122-3_8
Refereed	Yes
Divisions	Science > School of Mathematical, Physical and Computational Sciences > Department of Computer Science
Publisher	Springer
Download/View statistics	View download statistics for this item