Summer school

Project name

Fish fins morphology evaluation using CNN

Leader

Ing. Jan Urban, Ph.D.

Description of the activities of the project

• Fish fin morphology is significant measure in the context of damage and its implications for fish welfare. Damage to fins, whether through nipping, tearing, or fin erosion, can serve as a visible indicator of stress, aggression, or poor environmental conditions. It may compromise the fish's ability to swim efficiently, impacting its navigational skills and making it more vulnerable to predators. Moreover, damaged fins can hinder normal feeding behavior, reproduction, and social interactions, further affecting the fish's overall health and quality of life. Monitoring and addressing fin damage are critical aspects of ensuring the welfare of captive fish, as it allows for early identification of potential issues, prompt intervention, and the creation of conditions that support optimal fin health and functionality. Image analysis provides instruments for fin damage and anomalies detection. Advanced methods like machine learning enable algorithms to learn and differentiate between healthy and damaged fins. Automatised approach enhances the efficiency of fin health monitoring in aquatic environments, allowing for early detection, intervention, and improved overall welfare for captive fish populations.
• The aim of this project is to setup the experiments and train simple neural network for fish fin detection, segmentation, and classification.
• During the project, the student(s) will learn basic and related parts about data acquisition, image processing, annotation, machine learning, and neural networks training,

Students

• Number of students - 1-2
• secondary school students - NO
• university students - YES
• Prepositions (what should the students know): Basic algebra, basic algorithmization, basic statistics.