Cove Of Wonders

Cove Of Wonders aims to promote animal welfare while simultaneously educating a younger audience via an interactive and engaging learning experience while utilising A.I. to enhance functionality.

Comment

Inspiration

In the modern world, riddled with quick dopamine hits, with unparalleled access to technology, children are increasingly reliant on social media as their primary source of entertainment. As a result, children develop self-perpetuating bad habits and decreased attention spans, amongst other problems. On a different note, there are millions of wild animals in the world that suffer due to habitat loss, and a similarly massive number of domestic/feral animals that are victims of abandonment. We hope to solve both issues by providing a productive source of entertainment for children, while also educating them, and encouraging them to help participate in maintaining the wellbeing of animals.

Aim

Cove Of Wonders aims to promote animal welfare while simultaneously educating a younger audience via an interactive and engaging learning experience while utilising A.I. to enhance functionality. We have a vast array of features that provide a means for children to learn, while also encouraging them to take part in activities that benefit animals, benefiting their local communities. Since our target audience is mostly just children, their safety and privacy are our utmost priority, and creating a positive environment is critical.

Features

Quiz

Our AI-generated quiz uses a database to create an interesting and engaging experience, particularly for younger audiences. It has multiple game modes; a classic mode with 15 questions to answer, a streak mode where the quiz continues until the user makes a mistake, and a timed mode, where the user has only one minute to answer as many questions as possible. The questions progressively increase in difficulty, and users will be able to choose from a plethora of topics. Currently, we have one quiz related to animals and one which is general knowledge-based.

Forum

The forum feature aims to connect people with similar ambitions associated with animal welfare and interactive education, while also enabling users to share their ideas regarding such topics. The channels are carefully regulated to ensure a positive and healthy environment that nurtures growth. Additionally, it serves as a method to connect experts with children curious to learn (in a monitored area).

Gallery

The gallery depicts images of people helping animals and allows users to upload their own photos regarding the same. In the future, there will be points awarded for this. The images will be verified by developers while it is saved in a pending verification folder on a server. This serves a few purposes. Firstly, it ensures all images are relevant and appropriate. Secondly, it ensures images are unique. Lastly, it maintains user privacy, since images revealing personal details will be removed.

Artist's Corner

The artist's corner contains 4 different step-by-step tutorials that users can follow along. Each tutorial is of a different animal. There is a color palette, eraser and pen which the user can utilize to make colorful drawings. At the end of the tutorial, you can click on the green tick to save your image. Your completed images can be viewed in the Menu, in the profile page.

Trivia Corner

The under-development trivia corner contains interesting facts that would intrigue children and draw them to learn more about the topics provided.

Achievements

The achievements provide more incentive for the children to complete education-related tasks. There are currently 3 achievements connected with each quiz game mode. Each of them has two tiers that are dependent on the extent of the achievement.

Feedback

The feedback button links to a form that allows users to provide a rating for the app, report bugs, and also provide general feedback for the developers.

How we built it

Fundamentally, the quizzes are built using an OpenAI API (gpt-3.5-turbo), that was integrated with an information database (to improve the accuracy). It uses UnityWebRequest for the connection of the quiz UI in Unity with the server. The difficulty progression is managed via the increaseDifficulty boolean. This variable depends on whether the previous answer was correct or not. Repetition in questions is avoided by saving a the question-texts as a list and ensuring the new response does not match any previous questions. The topics are saved using PlayerPrefs from the dropdown in the quiz - menu and added to the prompt with the contextText string. The gallery uses NativeFilePicker to input files from the user, and upload them to the server that was built using node.js and hosted on railway. The gallery server has a password-protected, developers-only page that enables the verification of files pending approval. From there, the images are sent to an uploads folder, and served to the Unity client and loaded onto the gallery. The forum was included with the help of Vivox, and the artist's corner was [tbd]. The trivia corner is being developed entirely using Unity and C#. Audio was added via an audio-controller game object that had an audio-source component attached to it. Original textures were created using Adobe Photoshop and Paint.net, and others were imported from the internet. Animations and video assets were created using Premiere Pro and After Effects. Almost every part of the app was built using the Unity framework and scripted with C#. The backend/serverside development used node.js, a bit of html and css, and some json for serving image data to unity.

Challenges we ran into

Initially, we were using Photon for the networking part of the project, but it simply wasn't suitable for the purpose. There were many issues with the server rooms, and it was not efficient on memory and computing power. Eventually, we replaced PhotonUnityNetworking with a node.js server, and used Railway to host it. The railway command line interface was a bit problematic to use initially, but we were able to implement the server successfully. Another hurdle we faced was gpt-3.5-turbo's occasionally inaccurate responses. This was quite difficult to fix, and required us to dabble in a bit of prompt engineering. Combined with some trial and error and extensive bug fixing, we were able to get consistently accurate results. Occasionally, chat gpt's response format is inaccurate, and that was result in a runtime error. We fixed that by looping the DisplayNewQuestion function every time such an error occurred, and setting the upper bound the total tries to a 100, to prevent unnecessary depreciation in credits. The optimisation and efficiency of our program was initially quite poor. It was highly demanding on the ram and the gpu load was high. Fixing this required us to manually delete a lot of scripts and replace them with more compact and efficient code. We also got rid of dependencies on libraries and tried using simpler solutions instead.

Accomplishments that we're proud of

This was our first time combining different platforms, and we were complete beginners in node.js. Getting the entire server setup in place was quite fulfilling and the whole gallery looks quite nice. We believe it will prove to be quite beneficial to the public, and enable children to be more active and valuable members of their community, while also promoting animal welfare. The AI-Generated quiz also works spectacularly well, and the results seemed like a culmination of our efforts in prompt engineering. The feature is quite refined, works as an engaging platform for children to learn more across various domains, particularly relating to animals and zoology. Overall, we're quite happy with the way the UI turned out, and we feel the app is quite polished, with beautiful design, animations, effects, and music. It has a lot more attention to detail than our previous projects, owing in part to our rigorous and systematic bug-fixing. The app is also highly optimised, and we were able to maximise its efficiency by systematically replacing resource-intensive elements of the screen. Clever usage of scripting and game objects allowed us to limit the number of scenes, further reducing the RAM usage.

Impact

Cove of Wonders transforms screen time into a positive, educational experience that encourages children to care for animals and their communities. By combining interactive quizzes, creative tutorials, and a supportive online forum, the app fosters empathy, responsibility, and community involvement. It offers a constructive alternative to unhealthy digital habits, empowering children to make a real-world impact while learning and having fun. Through its focus on animal welfare, Cove of Wonders not only entertains but also inspires a new generation to take meaningful action in their local communities.

What's next for Cove Of Wonders

We aim to bring several new features to our app in the near future. Implementing a credit system for the number of animals helped by taking metadata along with the images in the gallery scene is a big one. This credit system will serve as a basis for a competition amongst users, with special awards being provided to winners. Since this would increase the incentive to take steps in promoting animal welfare, the feature would be immensely advantageous for the over-arching goal of the app. We are also planning to add more drawing tutorials, providing children with a more diverse set of drawing options, and developing their interest in animals and zoology. These tutorials would also be labelled, and teach users basic anatomy (of the animals they're drawing along the way). Integrating a computational neural network coded in python and integrated with unity using barracuda is another objective we have relating to this feature. Currently, we have 3 achievements and 2 tiers. Adding more achievements for completing drawings, and individual achievements for gallery uploads is the logical next step for them. We're also going to add more tiers for each of the existing achievements. Lastly, we're currently developing a trivia section that would provide interesting facts to users, particularly children, enhancing their interest in different topics. We're also going to divide the section based on topics, and put in images to make it more appealing.

Built With

Share this project:

Updates