Student: Jason W. Kollars, Faculty Mentor: Syed Hassan Ahmed Department of Computer Science, Georgia Southern University, Statesboro, GA 30460. Email:,

The lionfish is a predatory animal that has seen a dramatic increase in population around the world as an invasive species and local marine ecosystems become a threatened due to the lack of natural predators [1]. To combat this issue, unmanned submersibles are currently being developed to autonomously hunt such species. These robots are trained to detect lionfish by Computer Vision and Machine Learning (ML) algorithms [2]. However, we still face latency issues when it comes to the detection of live images in an underwater environment. We are actively studying and analyzing various methodologies to bring robustness to the detection schemes. In this abstract, we are evaluating the existing ML algorithms as well as provide a public Lionfish image dataset for others to train their machines. First, we make a training and testing image dataset of lionfish using an automated search tool. Then we feed the algorithms the training dataset where a trained model is produced. Using the separate testing dataset, we can validate lionfish image recognition. Learning models such as the TensorFlow [3] ML environment will be used [4], but others will also be evaluated for image processing along with an analysis for use. The output of our work is expected to show that with an online image dataset, machine learning algorithms can identify a lionfish and also such an approach can benefit other research efforts in ML-based detection and recognition schemes. References: [1] US Department of Commerce, and National Oceanic and Atmospheric Administration. “NOAA National Ocean Service Education: Lionfish Discovery Story.” NOAA's National Ocean Service, 6 Jul 2017. [2] Lombardi, Joseph, et al. “Autonomous Lionfish Harvester.” WPI Electronic Projects, 30 April 2018. [3] “TensorFlow.” TensorFlow, [Last Accessed: 12/4/2018] [4] Furlan, Carmelo, and Andrew Boniface. “Senior Project: Lionfish Detection System.” Cal Poly Digital Commons, 15 Jun 2018.

Additional Abstract Information

Presenter: Jason Kollars

Institution: Georgia Southern University

Type: Poster

Subject: Computer Science

Status: Approved

Time and Location

Session: Poster 3
Date/Time: Thu 2:15pm-3:15pm
Location: Student Recreation and Activity Center - Tripod 30 Side A