LEGO EMMET B200 Object Detection Computer Vision Project

LEGO EMMET B200 Object Detection Project

Project Overview

The LEGO EMMET B200 Object Detection Project is focused on developing a sophisticated object detection model specifically tailored for identifying LEGO bricks. The goal is to utilize this model to suggest potential LEGO builds based on detected bricks. The dataset used for this project, sourced from Kaggle, comprises highly realistic, synthetic images designed to closely mimic real-world LEGO bricks. This dataset contains 800,000 total images, featuring 200 of the most popular LEGO parts, with 4,000 images per part, all in 64x64 RGB format.

Descriptions of Each Class Type

The dataset consists of 200 distinct LEGO parts, each representing a unique class within the model. These classes include but are not limited to:

• Basic Bricks: Standard LEGO pieces of varying sizes, including 1x1, 2x2, and 2x4 bricks.
• Plates: Thin, flat pieces often used as a base layer in builds.
• Tiles: Smooth, flat pieces typically used to create finished surfaces.
• Slopes: Angled bricks used to add incline or decline in a build.
• Wheels and Axles: Components used for creating movable parts in LEGO constructions.
• Minifigure Parts: Components of LEGO minifigures, including heads, torsos, and legs.

Each class is vital for the model to accurately detect and suggest creative builds using the identified bricks.

Current Status and Timeline

• Dataset Preparation: Completed. The dataset from Kaggle has been preprocessed and is ready for model training.
• Model Development: In progress. Initial models are being developed and fine-tuned to improve accuracy in detecting LEGO parts.
• Testing and Evaluation: Pending. Once the model reaches a satisfactory level of performance, it will undergo rigorous testing to ensure accuracy and reliability.
• Deployment: Future phase. The model will be deployed in an application where users can upload images of LEGO parts and receive build suggestions.

Timeline:

• Q3 2024: Complete model development and begin testing.
• Q4 2024: Finalize model and deploy in a user-friendly application.

Contribution and Labeling Guidelines

• Contribution: Contributors are welcome to assist in model development, data augmentation, and improving detection accuracy. Please follow the standard guidelines for coding practices, ensuring your code is well-documented and adheres to the project’s coding standards.
• Labeling Guidelines: When contributing to data labeling or correction, ensure that each LEGO part is accurately labeled according to its corresponding class. Labels should be precise and follow the standardized naming conventions as per the class descriptions.