Explore the innovative AI and ML tools I've created to automate various tasks. From reading newspapers to searching for properties in receivership, these projects are designed to make life easier and more efficient.
Deep Learning-Based Anomaly Detection on Sidewalks
Deep Learning
The Anomaly Detection on Sidewalks project is a deep learning-based system designed to identify and classify non-pedestrian entities, such as bicycles, skates, and strollers, as well as anomalous walking patterns of pedestrians. Utilizing a dataset from UCSD, the project leverages advanced machine learning and computer vision techniques to preprocess and annotate data, ensuring the model accurately detects subtle anomalies in varying pedestrian densities and dynamic backgrounds.
The model undergoes rigorous training, starting with initial sanity tests to confirm the network components' functionality, followed by learning rate tuning to optimize performance. The evaluation process includes metrics like accuracy, precision, and recall, and incorporates an explainability library to understand the model's decision-making process.
The goal of the project is to automate the detection of sidewalk anomalies, thereby enhancing public safety. The model achieved impressive results, with an overall accuracy of 99.2%, precision of 99.65%, and recall of 97.5%, demonstrating its effectiveness in real-world applications.
Revolutionizing the way we read newspapers: An AI-powered Automatic Newspaper Reading System
Machine Learning, Computer Vision, Optical Character Recognition
The Automatic Newspaper Reading project is a machine learning-based system that can scan newspapers, identify and extract relevant ads using an object detection model, and perform OCR on the text to extract it from the image. The project utilizes state-of-the-art techniques in machine learning and computer vision to effectively process and extract information from newspapers with high accuracy. The end goal of the project is to automate the process of reading newspapers and make the information more accessible and readable for users.
Streamlining Literature Surveys: An AI-Powered Tool for Automatic Analysis of News and Research Literature
Natural Language Processing, Literature Analysis
The Automatic Literature Survey tool is an NLP-based system that can analyze news and research literature publications. The tool uses natural language processing techniques to extract and process relevant information from a wide range of literature sources, making it easier for researchers and experts to stay up-to-date with the latest developments in their field. The goal of the project is to automate the process of literature survey and help researchers to save time and effort while staying informed of the latest research and news.
Unlocking Opportunities in Asset Receivership: A Search Engine for Commercial, Estate, and Industrial Properties
Search Engine, Data Extraction
The Search Engine for Asset Receivership and Real Estate Ads is a tool that allows users to easily find commercial, estate and industrial properties that are in receivership. The engine utilizes data extracted from newspaper ads using the Automatic Newspaper Reading project and combines it with location data from Google Maps to provide detailed information about the properties and their receivership status. The goal of the project is to provide a user-friendly and efficient way for users to search for properties in receivership and help them to find the perfect opportunity for their needs.
The Mouse Logger project is a tool that allows for the passive collection of mouse movements during a questionnaire using JavaScript code and display the results. The project utilizes JavaScript and web development techniques to track and record mouse movements, providing valuable insights into how users interact with the questionnaire. The end goal of the project is to improve the user experience by understanding their behavior and making necessary adjustments to the questionnaire design.
Puzzles and Engaging Games: Two Smalltalk-based Projects for 8 Queens and War Card Game
Smalltalk, Game Development, Graphic Interface
The 8 Queens puzzle and War card game projects are tools that utilize Smalltalk programming language to develop graphic interfaces and code. The 8 Queens project allows for the placement of eight chess queens on an 8x8 chessboard without any two queens threatening each other, while the War card game allows players to choose the number of players and customize their deck. Both projects demonstrate the power of Smalltalk in solving complex problems and providing a visual representation of the solution. The end goal of these projects is to showcase the capabilities of Smalltalk in creating interactive and engaging applications.