Introduction to MyVision
MyVision is a user-friendly, free online tool designed to assist users in annotating images for generating machine learning training data specifically for computer vision applications. This innovative software aims to streamline the labeling process, making it both efficient and effective for handling extensive datasets.
Key Features
Drawing Tools
MyVision provides users with the ability to draw bounding boxes and polygons to label objects within images. These tools allow for precise identification and categorization of various elements, crucial for creating detailed training datasets for machine learning.
Polygon Manipulation
Enhancing its drawing capabilities, MyVision offers advanced features for polygon manipulation. Users can easily edit existing points, remove unnecessary ones, or add new points to refine their annotations. This flexibility ensures accuracy and adaptability in image labelling.
Automated Annotation
Annotating images can be a labor-intensive task. To simplify this, MyVision incorporates a pre-trained machine learning model called 'COCO-SSD'. This model automatically generates bounding boxes around objects, significantly reducing manual workload while maintaining data privacy by processing images locally on the user's browser.
Import and Conversion
For users with existing annotation projects, MyVision allows for seamless import of these projects into the platform. Additionally, it facilitates dataset conversion, making it easier to transfer annotations from one format to another without losing valuable data.
Supported Languages
MyVision supports both English and Chinese (Mandarin), catering to a diverse, global user base.
Getting Started
One of the conveniences of MyVision is its easy setup—no installation is required to use the application. Simply opening the index.html file gets you started immediately. For those interested in contributing to the project, setting up a local development environment requires Node version 10+ and NPM version 6+, and involves a few straightforward steps to install dependencies and run the project in watch mode.
For Contributors
For individuals looking to contribute to MyVision's development, all changes should be made in the src directory, with outcomes observed in the publicDev folder.
Citation
If you need to cite MyVision in your work, use the following reference:
@MISC{MyVision,
author = {Ovidijus Parsiunas},
title = {{MyVision}},
howpublished = {\url{https://github.com/OvidijusParsiunas/myvision}},
year = {2019},
}
MyVision stands out by offering seamless image annotation capabilities paired with advanced machine learning integration, suitable for both novice and expert users aiming to generate robust computer vision training data.