Proposal for a Local Database: Segmentation and Classification Algorithm

Abstract

The management of fisheries in Senegal presents many difficulties, at the level of the fishing docks the fish caught are neither declared nor collected for stock recovery and fish classification. To achieve the objectives of detection, segmentation is one of the most important aspects. It is in this context that segmentation has been the subject of several research themes in recent years. In the field of fisheries in Senegal, data collection is very difficult due to the fact that the techniques used are very often manual. Consequently, a local database adapted to the objectives of the survey is lacking. In this paper, we have proposed a semantic segmentation algorithm that tends towards intelligent systems for collecting artisanal fisheries catches. Data are collected by taking images of caught fish and on Fishbase. The collected data applied to the algorithm allowed us to obtain a set of segmented data with the masks of the images which constitutes our local database.