AI GitHub
Ultralytics for training and exporting YOLO vision models
Train, evaluate and deploy supported YOLO tasks with a common Python and command-line interface.
Key Takeaways
- The package provides task-specific pipelines for vision workflows such as detection and segmentation.
- Select a checkpoint for the intended task and prepare a matching labeled dataset.
- Fine-tune a small checkpoint and compare the original and exported models on held-out images.
- Training code and model distributions have licensing terms to review.
How it works
The package provides task-specific pipelines for vision workflows such as detection and segmentation. It supports model export paths for several deployment environments.
What you need
Select a checkpoint for the intended task and prepare a matching labeled dataset. Export after the preprocessing and output format have been tested in the original runtime.
What to try first
Fine-tune a small checkpoint and compare the original and exported models on held-out images. Check bounding-box coordinates and confidence thresholds after export.
Test the exported model on a scene with no target objects. It is a useful guard against discrepancies in confidence filtering or output decoding.
Where to look for failures
Training code and model distributions have licensing terms to review. Verify preprocessing and postprocessing parity for the target runtime.
Related Franklin pages
Choose between text-guided boxes, prompted masks and downstream tracking. Preserve a few ambiguous images and videos so each stage can be evaluated independently.
- Generate object masks from visual prompts and propagate them through video. SAM 2 for prompted image and video segmentation
- Find image regions that match natural-language phrases. Grounding DINO for text-guided object detection
For an overview of the types of predictions vision systems produce, see Franklin’s glossary. Computer vision
Project documentation
Try it for your workflow
Explore the project’s documentation for setup and usage, or save this page to a collection for later.
Read the project documentation ↗