 The next presentation is by Amos Barkay on integrating electronic monitoring and reporting technology for fisheries management and compliance and exploring the variability of deploying official technology independent of human interventions related to the monitoring and surveillance of commercial fishing operations. Good day. My name is Amos Barkay. I am the CEO and co-founder of OLSPS. Today, I will be presenting to you our work on the integration of our existing electronic reporting technology, named OLRAC, with third-party onboard electronic monitoring system for fisheries management and compliance. I will also share with you our vision regarding the incorporation of AI technology as part of our envidious electronic monitoring and reporting integrated solution. OLSPS was founded in 1989 by Dr. Mai Bergen myself for the purpose of providing commercial fishing operators with analytical insight and with related software products needed for the quantitative management of fish resources. The photo in this slide was taken last year, just shortly before the COVID pandemic changed our life maybe forever. Presently, OLSPS employs 40 vibrant and diverse individuals, most of whom are highly qualified and professional in the field of software design, fishery science, mathematics, statistics, data modeling, database design, and more. The entire company is now cloud-based and operating remotely with each other and with all our clients around the world. In recent years, there has been an increasing demand for title control of commercial fishing operations in national and international waters. In addition, public pressure was mounting for the inclusion of ecosystem consideration in commercial fishing regulation and management targets. To address this demand, fisheries managers needed many more and far more reliable data from fishing operations. These needs were translated into a tsunami of fishing regulation and data reporting requirement. On-board human observers were initially viewed as a significant way to improve both data collection and real-time on-board monitoring. However, the deployment of on-board fishing observers was found to be very expensive to deploy and difficult to manage and often surmount to a relatively low operational coverage. One obvious solution for the greater demand for a large amount of reliable data was the introduction of modern data technology to the commercial fishing sector. OLSPS has developed a highly sophisticated yet simple-to-use ELOG software system named OLRAC. The OLRAC ELOG was designed to be infinitely customizable for the needs of any fisheries in any countries using any fishing methods. The OLRAC ELOG can also be easily modified to address ever-changing data requirements. This comprehensive data management system is comprised of two units, the OLRAC Dynamic Data Logger, which is the vessel unit, and the OLRAC Dynamic Data Manager, which is the web-based fleet management server. The OLRAC DDL ELOG software is capable of tracking vessel movements in real-time while collecting and recording many types of data, including catch, gear, discard, wildlife interaction, environmental data, and more. Data can be recorded and stored in any kind of format and can include video clips and images. The OLRAC DDL web server can manage data from an entire fleet of vessels to the level of a whole fishery or the entire national fishing fleet. The OLRAC system is presently deployed in many fisheries and is used daily by hundreds of vessels around the world. Last year, OLSPS initiated a joint venture project called MRAP, stand for Electronic Monitoring and Reporting, funded by EA Grants. OLSPS' proposition for this project was to create a tightly integrated electronic monitoring and reporting system. The idea of our proposed system was that the ELOG's reports verification process will be automated to allow the easy extraction of the relevant media using data and time data as a common key between images and report. As part of the integration project, OLSPS offered to test the viability of using AI image recognition technology to automatically scrutinize, EMProduce images and to extract key information such as catch by species estimation and key events like fishing gear deployment and holding, the forbidding discarding of fish, wildlife interaction, etc. This was all the need for manual inspection of the EM footage. My technical team quickly established that image recognition technology is widely available through various open source platforms which we can use and integrate in our system with ease. However, all these AI software tools require very significant amounts, sometimes many thousands, of label relevant images in order to produce accurate result. This requirement presented us with a significant challenge as it became evident to us very quickly that the acquisition of huge number of label images from self-party sources is virtually impossible. In order to try and tackle this challenge, we decided to make the data labeling and model training part of our integrated electronic monitoring and reporting system. This will allow a layperson, user of our software, to conduct a labeling and model training work alone over an extended period of time during normal operation. We are currently investigating few possible models, namely semi-supervised learning with different levels of human intervention. Thank you for allowing me the opportunity to present our technology and vision to you. I'm happy to answer any questions you may have or alternatively please feel free to contact me via my details captured below. Thank you very much for sharing the vision and the work of your team, Amos. I was just thinking out loud here when you're talking about your labeled image data sets and the challenges everyone has to get to these labeled image data sets and how we can think of incentivizing a kind of global repository of labeled data sets which have had the oversight of, similarly to how we put books together at FAO, these ID books we work with museums of the world and so on. Do you see there's an opportunity there to help you develop these type of fully autonomous systems, to bring in other players to make this a reality so that we can work more hand in hand? I notice even your team seem to be coming from South Africa and Israel and so on, so multinational teams. How do we bring a multinational vision to the requirement to build really good labeled data sets for everyone to be able to use? Have you got any ideas on that, Amos? Yeah, thanks. Thanks everybody. Good morning. I have many ideas, but I'm not sure that I have many solutions. It's extremely challenging. I'm actually presently involved in another project funded by the European Space Agency about data security and data sharing and how to bridge the gap between data protection and the wanting to share data for scientific, commercial compliance reasons. And as part of it, we approach many organizations that we're working with, Fisherman Organization, Authorities Organization, and right now I can tell you that 80% to 90% of the time the response is negative. People are not prepared to share data. It's even difficult. I mean, it's challenging enough to convince them to have come around the vessel as the notion that some of the data that we are collecting will be shared even in a completely protected, secure, anomalous way is freaking them out for all sorts of reasons. Most of them, to my mind, are not justified. It's a matter of education. It's a matter of continuous challenge. And it's maybe a matter of some type of national and international convention and agreements, but to go to a Fisher or to go even to a fishing association, a fishing company, and to tell them that you want or you need to share their data is almost like actually cutting yourself from any business relationship with them right now. So it's extremely challenging. That talks very much to the challenge ahead of us. And I think that's where sometimes, for example, we have the Committee of Fisheries on Fisheries at FAO where the world's governments come together and they make deals around the things that they believe they need and they normally bring finance to those types of problems. So these are the kind of, you know, when you do have a problem, you can't just chuck money in it, but maybe we can turn to Anton Ellenbrook who can give us some ideas on how FAO has dealt with these types of questions before to build these collaborations around recognized need by the global community. Anton. So for those of you already know FAO, so we hold several secretariats with where we work with fisheries organizations around the planet to start discussing, for instance, standards and guidelines for fisheries and aquaculture. So that can be seem very trivial, but for instance, to have a global definition of what is a fishing net and what classifications there are in fishing nets. It is, of course, it will have implications if you start building a regulation around your fishery. So for FAO, I think it's essential that we can talk to all stakeholders, so from consumers all the way up to national governments, but we need, I think, we cannot talk to all individuals on the planet, so we need to work with you using a structural approach. And I think through the current setup where we talk to fisheries and aquaculture, national international organizations, I think that is also something where we need to bring in these topics surrounding AI, and especially for them, I think for FAO, more the international aspects of using AI, so from ethical aspects to more data quality and data standards, data sharing aspects. But I think in, as an organization, we have this opportunity that we can directly talk to a lot of governments and international organizations. Well, Amos, I think these are the kind of, you know, these are the kinds of initiatives that we're trying to put together now to build a momentum to those types of meetings that will be held and so on. And so it's going to be important for us to make sure we have a platform where these issues can be discussed and letters of support for approaches can come not just from governments, but also from major players and major research groups who are looking for this type of need. Because often when you do put in these types of requirements, there's a financial need. And to get that financial need delivered, often it takes people saying, yes, we need this. This is not something that's come out of somebody's theory of what's needed. This is actually needed. So it'll be very valuable in the future for us to leverage the community that we're building through these types of meetings.