According to the Food and Agriculture Organization of the United Nations (FAO), between 1974 and 2017, the proportion of fish stocks found at biologically sustainable levels dropped from 90% to about 65,8%, on average worldwide. At the same time, globally, between 1976 and 2018, the value of fish exports increased from 7,8 billion dollars to about 164 billion dollars. This increase in the value and demand for fishing makes inspection even more necessary, because from illegal fishing, the products generated by it reach the foreign market, harming local markets, increasing the risk of poverty. According to Telesetsky(2015), the value generated by INN fishing is somewhere between 10 billion and 23,5 billion dollars per year, based on a US estimate in 2011.
INN Fishing is an acronym for Illegal, Unreported and Unregulated Fishing. Where in illegal fishing is the act of fishing using illegal means or places of preservation/prohibited, fishing does not declare is the act of fishing without proper declaration of the means used, fishing locations and information relevant to the law, unregulated fishing it concerns fishing done in places with no fishing regulation, where there is no fishing control.
Currently, Brazil does not have concrete data on fisheries in general, as one of the last reports on this was issued in 2011. This ends up blinding, statistically, both the fishing community, as well as environmentalists and entities that protect the aquifer ecosystem. Along with the lack of information, we have an ineffectiveness in preventing IUU fishing, given the vast extension of the Brazilian coast, which allows the practice in preservation areas, which are prohibited for fishing for various reasons. In some places, the environmental military police already use drones to prevent this practice, however, their monitoring depends on an agent monitoring this inspection, which ends up reducing the effectiveness and practicality of monitoring. With this in mind, to help monitor and prevent illegal fishing, an application based on Python, Machine Learning and Image Processing was developed, where, from a video coming from a drone, we can identify and demarcate the presence of a vessel in a prohibited and/or monitored area, generating a report with the image of the vessel and information about its encounter for analysis by the responsible authorities. For the development of this project, databases relating to different images of vessels and water bodies of different types were used, representing what a vessel is and what it is not, for the training of artificial intelligence. Using artificial intelligence to screen static images, detecting vessels in test images. After screening, the report generator is developed, using the ReportLab library to manage the pdf files, where a function is created that is called after the detection of vessels by screening, generating a report with the image of the detected vessel. To carry out a detection and tracking, in the image, of a vessel, OpenCV was used, demarcating in the image where the vessel is. The results obtained so far show that the programmed functions perform their tasks, the artificial intelligence, together with the screening, present an accuracy of 97%, generating a report with the image detected by the screening. As for detection, tracking and demarcation by OpenCV, it performs its function, being able to demarcate more than one vessel in an image. For the future of the project, the implementation of automation of drones is planned, so that the system can carry out the inspection in a complete and autonomous way, facilitating and assisting in the inspection, monitoring and preservation of protected areas against fishing..