Image Analysis in Marine and Maritime domain using computer vision techniques

Research / Pillars & Groups / Visual Sciences / BIO-SCENT / Internships / Image Analysis in Marine and Maritime domain using computer vision techniques
16 January. 2021
Ocean imaging has become very popular nowadays, as an easy way to monitor the marine and maritime environment. Monitoring can be utilized for the protection of several threats that the Marine and Maritime industry faces or for the automation of everyday processes in the domains of border control, safety and security, fisheries control, customs, environment, defense, etc. The manual analysis of the large amounts of monitoring data is very time consuming and difficult task. Thus, the computer vision techniques have been extensively used to automate the process. Although computer vision has many advances the last years, the research for marine and maritime is at an early stage and several challenges remain open. 
Required Skills
1. Computer Programming Skills 
Skills Level
Through the use of computer vision, the integrated ocean monitoring can be achieved offering ways to exchange and visualize information and data. Although some techniques for automated detection, identification, measurement, tracking, and counting objects (e.g. fishes, oil spills, etc.) in underwater or sea surface optical data streams have been proposed, very few automated systems are available which incorporate all appropriate techniques for highly successful and accurate results. The main objective of this project is to develop beyond the state-of-the-art technologies for analyzing ocean visual data. For this purpose, an extensively research review will be conducted to define the challenges of computer vision techniques in the domain, and then innovative techniques will be developed to tackle them which may include: pre-processing methods for enhancing the image quality, image segmentation to extract and locate the visual content, feature extraction and machine learning for modelling the object categories, and innovate image classification schemes using deep learning techniques.