OvidijusParsiunas/myvision
Computer vision based ML training data generation tool :rocket:
What it solves
MyVision provides a free, browser-based tool for creating the training data needed for computer vision machine learning models. It simplifies the process of manually labeling images by providing tools for drawing bounding boxes and polygons, as well as automating parts of the process.
How it works
Users can manually annotate images using bounding boxes and polygons. To speed up the process, the tool integrates the COCO-SSD pre-trained model to automatically generate bounding boxes locally in the browser, ensuring data privacy. It also supports importing existing projects to continue labeling or to convert datasets between different formats.
Who it’s for
It is designed for developers and researchers who need to generate high-quality labeled datasets for computer vision tasks.
Highlights
- Automated Labeling: Uses the COCO-SSD model to automatically suggest bounding boxes.
- Polygon Manipulation: Includes advanced tools to edit, remove, and add points to polygon labels.
- Format Conversion: Allows importing and exporting datasets in various supported formats.
- Browser-Based: Operates locally in the browser, meaning no complex setup is required to run the basic tool.