Midshield - [Blockchain + Security] & [AI Image Analysis]

💡 Inspiration

A lot of notable crime cases go unreported because they are observed by passive bystanders. We want to incentivize people to report crimes and their evidence to authorities. Additionally, we want to make it easier for detectives and officers to manage evidence. With our tool, the writing of the report is automated, allowing officers to focus on more important tasks. We also wanted to help officers find similarities between images that they might not catch on their own. That's why we added the comparison feature to allow users to compare multiple crime scenes.

🖼️ What it does

This app allows users to upload images of crime scenes. To preserve the integrity of these images, they are securely stored on the blockchain. Users can also submit reports and attach relevant images, which are automatically formatted into a comprehensive report using AI. Law enforcement officers can compare multiple images using AI-driven image recognition to identify patterns or similarities that might not be immediately apparent. Furthermore, by integrating blockchain technology, users who report crimes can be rewarded with cryptocurrency for their contributions, providing an incentive for active participation in public safety.

Crypto rewards send on main net.
Metamask wallet support.
Automatically writes a report for crime scenes.
AI allows users to compare different crime scenes and writes a similarity report on them.
Stores user submissions.
google authentication.
Image hashing to ensure images have not been tampered with
Subgraph implementation
Subgraph data display

🛠️ How we built it

We built it using Next.js for both the frontend and backend. Supabase was utilized for Google authentication, blob storage, and managing the Postgres database. We used Drizzle ORM for database interactions, and TypeScript for robust and type-safe code. For wallet authentication and transactions on the blockchain, we integrated the Metamask SDK. The Graph was employed for data visualization and querying, enabling us to efficiently access and present blockchain data. Finally, we leveraged the OpenAI API for advanced image recognition and comparison, allowing our app to analyze and compare crime scene images with high accuracy.

🚧 Challenges we ran into

Setting up reward transactions securely and efficiently on the blockchain was a significant challenge.
Deploying the subgraph using The Graph presented some technical difficulties that required in-depth troubleshooting.
Image recognition using GPT-4 required fine-tuning to ensure the AI could accurately analyze and compare crime scene images without error.

🏆 Accomplishments that we're proud of

Successfully deploying the subgraph in record time, which was a critical component of our project.
Overcoming the learning curve associated with smart contract development and blockchain transactions.
Integrating advanced AI image recognition into our app, which added a significant layer of functionality and utility.

📚 What we learned

Before this project, we had little experience with smart contracts or blockchain transactions. By the end, we not only mastered these skills but also gained a deep understanding of the advantages of using an indexing protocol like The Graph to access and interact with blockchain data efficiently. Additionally, we learned how to leverage AI to enhance the capabilities of our application, particularly in the domain of image analysis and comparison.

🚀 What's next for Midshield - [Blockchain + Security] & [AI Image Analysis]

Moving forward, we plan to refine our image recognition and comparison algorithms further to improve accuracy and efficiency. We also aim to expand the reward system to include more cryptocurrencies and potentially other forms of rewards, such as NFTs for active users. Enhancing user experience and making the app more accessible to a broader audience, including international law enforcement agencies, is also a priority. Lastly, we are exploring the possibility of integrating real-time crime alert systems and predictive analytics to proactively prevent crime and enhance public safety.