"Draw Sense: A handmade Machine Learning Web Project crafted from the ground up. This initiative encompasses the entire journey, from meticulously collecting and creating the dataset to manual classification using JavaScript. The project, named 'Draw Sense,' focuses on detecting hand-drawn doodles, classifying them into eight labels: car, fish, clock, tree, bicycle, pencil, guitar, and house. The dataset is enriched with four features—width, height, path count, and point count—whereas the actual classification employs only width and height. For the classification task, a K-Nearest Neighbors (KNN) model is implemented in JavaScript, eschewing the use of libraries for a truly independent creation. While the absence of KD-trees impacts efficiency, the model performs admirably, achieving a final accuracy of 54.6 percent. 'Draw Sense' showcases the potential of a hands-on approach to machine learning, proving that even without conventional tools, a functional and insightful project can be realized."