Bubble Answer Computer Vision Project

Dataset Overview

This dataset contains images of individual bubble answers extracted from various answer sheets commonly used in Indonesian schools. The dataset is designed to aid in the development and testing of AI models for automated bubble sheet grading. Each image is categorized into one of four types, representing different ways students might mark their answers. Bubble Types

1. Default: The bubble is unmarked and remains in its original state.
2. Filled: The bubble is completely filled in, indicating a selected answer.
3. Crossed: The bubble has a cross mark, which is another common way students indicate their answer.
4. Invalid: The bubble is marked in a way that does not conform to standard marking practices, such as being partially filled or having other extraneous marks.

Usage

This dataset is suitable for training and evaluating machine learning models for tasks such as image classification, optical mark recognition (OMR), and automated grading systems. By providing diverse examples of bubble markings, the dataset helps in creating robust models capable of handling various answer sheet formats used in Indonesia.

Example Applications

• Automated Grading Systems: Develop AI models to automate the grading of bubble sheet tests, saving time and reducing errors for teachers.
• Optical Mark Recognition: Enhance OMR systems to accurately interpret different marking styles on bubble sheets.
• Educational Tools: Create tools to assist educators in quickly and accurately processing test results.