Human activities are a threat to marine species that live on reefs all over the world, and many conservation efforts have been put in place to lessen the damage. Scientists lead successful citizen-based environmental projects where data is collected by volunteers, which gives them valuable information for their research.
For my graduation project for Data Driven Design Master's program, I've looked into how recreational divers can help monitor marine species by sharing pictures they take while diving and what keeps them engaged in this activity. The possible solution is a mobile app for divers that lets them connect with each other and, most importantly, allows to share pictures from underwater. Artificial intelligence is used to classify the species of photographed animals. This, along with data like location and date, gives an overall picture of how organisms in this biodiverse environment behave and how healthy they are.
.png)
Semi-structured interviews with users and expert interviews were done for this project, a survey was distributed to the target audience, and an application prototype was developed and evaluated. Expert interviews gave critical information about animal monitoring techniques and the need of involving citizen scientists in data collection. In addition, following preliminary research, a survey was distributed to divers to validate which elements are attracting users to join in the initiative. Based on the information gathered, a mobile application prototype was created and tested for further evaluation. Each user test was followed with co-reflection to acquire more insights into features that hooked users.
The expert interviews were conduced to better comprehend the significance of animal observations and to explore potential challenges. In May 2021, three interviews were performed with professionals with experience in animal monitoring, both underwater and on land. All of the experts are lecturers at various universities and work in the field with various investments, they were specialists in monitoring flora, amphibians, fish and mollusks in freshwater reservoirs and the Baltic Sea, and long-term reef monitoring in Okinawa, Japan.
The assumptions of the proposed solution were validated by experts. Because scientific monitoring does not need fishing, it is especially successful in protected areas. The more data collected, the better. They argue that because fish swim, precise GPS data are not always required. While a localized region may suffice, reliable data allows scientists to understand about the dynamics of individual species. Experts believe that this concept has the potential to play a significant role in dive destinations.
Throughout the preparation for the design of the application interface, semi-structured interviews with divers were conducted to acquire their perspectives, goals, experiences, and acknowledge the motives engaging them in the project. Divers had to have prior underwater photography experience.
Interviewees recognized the main disadvantages of currently existing apps and shared their ideas and preferences for prospective improvements. They claim that most applications are difficult to use and are just concerned with the process of diving logs. When asked what future features they would want to see, participants mentioned networking with friends, diving location descriptions and ratings, and activity rewards.
Scuba divers and free divers were the survey's intended audience. The online questionnaire consisted 21 questions concerning diving experience, citizen-based projects, incentives for participation, and components that encourage user engagement, including binomial, multiple choice, open, and Likert-type questions. To assure long-term participation, this strategy focuses on inquiries about the factors that drive user engagement (see Appendix B for the list of the survey questions).
Respondents were asked about their diving experiences and motives for participating in citizen-based initiatives. They were asked what would inspire them to join the initiative. The survey asked, "Are there any more features that would make you join the app?" Divers were asked about underwater cameras. Most responders (55.4%) bring a camera on dives; 37.5% do so sometimes. 28.6% of participants upload underwater photos regularly; 46.4% do so occasionally. They were also asked where they'd submit underwater photos so user channels might be investigated. 50% share it on Facebook and Instagram. Most (57.1%) are unaware with citizen science projects, in which people create study subjects, collect and analyze data, evaluate results, and develop new technologies. Altruistic (63%) and personal (33%) goals inspire participation, with a mix (3.8%).
The insights gathered during the interviews contributed in the development of the prototype's first stage. It was composed of two parts: a mobile application with a clickable interface and software with a machine learning model for image recognition. To explore, a wide range of specific technologies were available. Lobe, Scikit-image, TensorFlow, Keras, and Flask were among the approaches tested. Flask enables the deployment of a front-end web application in which a photo can be uploaded, sent to the model, which 19 analyses the image, and the prediction shown. To educate the model to recognize the species, the software described above requires a data set. A small dataset obtained from Kaggle.com, which included around 1500 photos of 14 different shark species, was utilized to train the model. Another critical aspect was retrieving EXIF data from images; because many cameras have built-in GPS systems, it is easy to identify the date and location of an image. The user validates the extracted data, but he is not forced to type it manually, lowering the potential of error.
While the software's primary focus was data processing, the application's interface was designed in two stages: designs on paper and a digital interface built with the Sketch app. Because divers travel frequently, the interface was created for a mobile platform. The purpose of the design was to test and analyze features that keep users engaged in the project.
Usability testing identify issues that people encounter while interacting with the prototype. A group of divers got tasks to perform within the paper prototype and software solution for the first iteration. The second iteration was carried out in the same manner, however, on the basis of the high-fidelity prototype and the modified software. They were all asked to think aloud, which provided insight into what consumers were thinking. The purpose was to identify problems that people had when using the prototype. After careful examination, the information gathered was used to evaluate the prototype.
A co-reflecting session with a user was held after each prototype usability test. The topic of the study, as well as potential solution elements, were addressed during the session. The discussions were then followed by questions like, "Which features are you interested in?" and "Would you like to draw attention to any aspects that are not included in the prototype but appear to be of importance to you?" The purpose of the co-reflection session was to reflect on the question "How to develop and maintain long-term diver engagement in citizen science initiative?" and also to obtain feedback from the target audience on the prototype in order to evaluate and improve the solution by collecting additional inputs for functionalities.
The goal of this project is to understand how certified divers may help marine monitoring by voluntarily sharing photographs from their dives, as well as what factors keep them involved in the program. My research looks at existing projects that include divers in supporting marine conservation efforts, the importance of citizen-generated data, and the integration of artificial intelligence and citizen science. Taking collected insights into consideration and developing them into existing application could result in a useful data collection tool that can be used in the long run.
