Soham Patil@sohampatil17

This project embodies the fusion of two technologies: image generation with DALL-E and video synthesis with Stable Diffusion XL Turbo. Our initiative is centered on producing meme GIFs, which have become an important part of digital expression. The process begins with the generation of static images using DALL-E, an AI model known for its ability to create detailed and diverse visuals from textual descriptions. These images serve as the foundation and context for the subsequent video synthesis. The Stable Diffusion XL ("SDXL") Turbo is deployed for transitioning from stillness to motion. This particular model, a specialized variant of the Stable Video Diffusion (SVD) Image-to-Video, is adept at generating short video clips using a still image as the conditioning frame. Trained to produce 25 frames at a resolution of 576x1024, the model ensures each sequence is not just fluid but also retains temporal consistency, thanks to its finetuning from the SVD Image-to-Video [14 frames] and the f8-decoder. The result of this process is a collection of meme GIFs that are not only humorous and relevant but also boast a high degree of originality and quality. With the support of Stability AI, this generative image-to-video model unlocks new potentials in content creation, offering meme enthusiasts (everyone!!!) and digital content creators a tool to engage and entertain their audiences.

WebAlly is an AI-powered code-writing software that helps developers and designers create ARIA-compliant websites, focusing on improving web accessibility for the disabled, particularly those with visual impairments. By simply pasting your webpage URL, WebAlly generates a detailed accessibility report for your website, an ARIA (Accessible Rich Internet Applications) compliance score , and updated code to implement those recommendations! The tool leverages GPT-4 to analyze website code (we use BeautifulSoup for scraping your website's content and code), identifying accessibility issues and calculating an ARIA compliance score on a 0-10 scale. This score assesses adherence to the W3C (World Wide Web Consortium’s) ARIA accessibility standards. WebAlly's not only pinpoints issues but also generate revised code using OpenAI, enabling developers to easily implement changes and make their websites more inclusive for users with disabilities. We use TrueLens text-to-text eval to log and get feedback for our LLM Key Functionalities: Website Scraping: Upon receiving a URL input from the user, the application extracts HTML and CSS content from the specified website. This process forms the basis for a thorough accessibility evaluation. Accessibility Report Generation: Using the scraped website content, the web app analyzes the site for potential accessibility issues. This function leverages the capabilities of a GPT-4 to interpret the website's content and identify areas that might hinder accessibility. ARIA Compliance Scoring: The application calculates an ARIA compliance score on a scale of 0 to 10. This score quantitatively reflects how well the website adheres to established accessibility standards, particularly those pertinent to creating accessible web applications.

SahayakAI is an AI-driven software tool designed to transform the way legal documents are translated and understood in India. Legal documents in India are predominantly in English, which presents a formidable challenge in legal understanding and literacy for the vast majority of the population. Recognizing that over 90% of the Indian population speaks languages other than English, with a majority of 340M+ being native Hindi speakers, SahayakAI seeks to bridge the significant language barrier in legal contexts. The existing solutions, such as human translators and generic translation software, have notable limitations ranging from being too expensive to or lack of understanding of legal terminologies. To solve this, I built an AI solution that translates English 'legalese' into Hindi within seconds! We use Cohere's state-of-the-art multilingual AI model Aya to achieve this. Aya's advanced understanding of both English and Hindi ensures that the translations provided by SahayakAI maintain the integrity and intent of the original legal documents. Language translation services is a $40B+ global market on the brink of disruption due to multilingual LLMs like Aya.