Rakshak

Rakshak website Dashboard
Rakshak website timeline video
Rakshak website timeline image
Rakshak website timeline text
Rakshak chrome extension

Inspiration

The inspiration behind the project stems from the growing need to address and combat the proliferation of inappropriate and harmful content on the internet. With the widespread usage of the internet, there has been an increase in the dissemination of explicit, offensive, and obscene content. Such content can have detrimental effects on individuals, particularly vulnerable populations such as children and victims of harassment or abuse. The aim of the project is to create a solution that proactively identifies and blocks obscene content in real time, ensuring a safer online environment for users. The motivation behind extending the detection capabilities to cover all types of media files, including text, images, and videos is to ensure that no avenue for inappropriate content goes unnoticed.

What it does

The real-time obscene content detection web extension utilizes advanced algorithms and machine learning techniques to provide seamless functionality across all major browsers. Once installed, the extension operates in the background, constantly monitoring the content being accessed by the user. It performs a comprehensive analysis of various media files, including text, images, and videos employing contextual understanding to identify explicit or offensive material. When a potentially obscene piece of content is detected, the extension promptly blocks its display, preventing the user from being exposed to harmful material. This proactive approach ensures a safer browsing experience, particularly for vulnerable individuals. The project also provides an analytics dashboard where obscene media can be monitored.

How we built it

Backend Services: Our backend services utilize FastAPI, a high-performance Python framework for building web APIs. It offers efficient handling of HTTP requests and supports asynchronous programming. TensorFlow, a popular deep learning framework, is used to develop models for text, image, and video analysis to detect explicit content. These models are optimized for fast inference using TensorFlow Lite (TFLite) and are deployed as separate services for text, image, and video processing. We leverage multiprocessing to process video frames in parallel, enhancing the speed of video inference.

MongoDB Atlas serves as our database solution for storing user interaction data and content metadata. It provides scalability, allowing us to handle increasing data loads while maintaining performance. With support for unstructured data and powerful sharding capabilities, MongoDB Atlas enables efficient storage, retrieval, and analysis of diverse content formats, contributing to the effectiveness of our Obscenity Blocker Solution.

Frontend Application: We have built a Chrome extension which is developed using JavaScript, a versatile programming language widely used for web development. It seamlessly integrates with popular web browsers, providing real-time content filtering capabilities. By scanning web page content and communicating with our backend services, the extension ensures a safe browsing experience by blocking explicit content and promoting a secure online environment for users.

Another is the analytics dashboard which is integrated into our solution and provides users with a comprehensive view of content filtering statistics, user interactions, and other relevant analytics. This empowers users to monitor and manage the system effectively, enabling informed decision-making and proactive measures against obscene content. The frontend application is developed using React.js, a JavaScript library known for creating interactive user interfaces. It follows a single-page application (SPA) architecture for seamless navigation without full-page reloads.

Deployment: For deployment, we leverage the power of Google Cloud Platform (GCP) by hosting our backend services on Google App Engine. This fully managed serverless platform offers automatic scaling and robust infrastructure, ensuring efficient handling of varying workload demands. By utilizing Google App Engine, we can focus on developing our application without worrying about infrastructure management. To enhance the performance, security, and manageability of our APIs, we integrate Apigee, an API management platform. Apigee acts as a proxy layer, abstracting the backend APIs and providing advanced features such as security controls, rate limiting, analytics, and a developer portal. This integration allows us to effectively manage and monitor our APIs, ensuring their reliability and performance.

Challenges we ran into

We came across a set of challenges while developing this project. The main issue was developing accurate and reliable algorithms for detecting obscene content is a complex task. The algorithms must be trained on diverse datasets and continuously updated to adapt to evolving trends and new forms of explicit content. Another challenge was to design the algorithms and infrastructure that must be scalable enough to handle the demands of a broad user base without compromising performance or introducing latency issues that can negatively impact the browsing experience.

Accomplishments that we're proud of

Our primary focus was to develop highly accurate and precise algorithms, and we are delighted to announce that we have achieved an impressive average accuracy rate of 93% across various media types. This accomplishment underscores the effectiveness of our solution in accurately identifying and segregating obscene content. By leveraging these algorithms, our browser extension seamlessly integrates with popular web browsers, ensuring the reliable detection and filtering of inappropriate or offensive material. Users can now browse the internet with confidence, knowing they are protected from encountering explicit content.

Furthermore, our solution offers an analytics dashboard that provides insightful analysis of the spread of obscene content over time. This comprehensive dashboard allows users to monitor and gain a deeper understanding of the prevalence and distribution of explicit materials. Our solution's combination of cutting-edge algorithms, a powerful browser extension, and an intuitive analytics dashboard establishes a robust and comprehensive approach to content moderation, ultimately fostering a safer and more secure online environment for all users.

What we learned

The project of real-time obscene content detection using a web extension across all browsers and all types of media files provides valuable insights and learnings for improving online safety and user experience. Through this project, the importance of algorithmic accuracy and continuous training on diverse datasets has been emphasized, highlighting the need for robust and adaptable content-filtering systems. The challenges of scalability and performance have underscored the significance of efficient infrastructure to handle high volumes of media files without compromising speed and user experience.

What's next for Rakshak

Looking ahead, our future plans include the development and implementation of an Android application to further expand the reach of our content moderation solution. By creating a user-friendly and efficient Android app, we aim to provide seamless access to our advanced algorithms and features, enabling users to protect themselves from obscene content while using their mobile devices.

In addition to the Android application, we are actively seeking partnerships with prominent browser developers, content platforms, and internet service providers. By collaborating with these industry leaders, we can integrate our real-time content detection web extension directly into their platforms. This integration will empower users of these platforms to enjoy the benefits of our solution without the need for any additional installations or configurations, ensuring a smooth and hassle-free experience.

Through these strategic partnerships and technological advancements, we envision a future where our content moderation solution becomes deeply embedded in the digital ecosystem. By reaching a wider audience through the Android application and partnering with key industry players, we aim to create a safer and more responsible online environment for all users, regardless of the platform or device they use.