Better machine learning models, fast

  

Machine learning models are only as good as the data that they are trained on – garbage in, garbage out. When it comes to getting your training data in the best shape possible, good annotations are key. When it comes to object detection in images, many architectures focus on bounding-box annotation. These low fidelity annotations include extra pixels of non-target information that often result in unintentional biases in your models. 

Instance segmentation has long been as the gold standard solution to these problems, however creating segmentation masks has historically been too slow, expensive and/or difficult to be a viable option for many. 

Zegami Image Annotator is designed to solve this problem: a visual, intuitive and fast way to curate training data sets, create segmentation masks and then integrate into your existing machine learning workflow. 

Image Annotator benefits

Pixel perfect segmentation

Most annotation tools are tedious to work with. Drawing boxes around thousands of objects makes it difficult to focus and the quality of your training data drops. With Image Annotator creating masks is straightforward: a simple box, a quick adjustment and you’re done. If you prefer drawing polygons or painting masks by hand, then you are free to do so.

Collaboration

Collaborate with Zegami Image Annotator to share the workload, peer review each other’s work and track overall progress of the project. Whether tracking annotations by the person that made them or by class, Zegami makes visualising your assets simple. 

Zegami integration 

Zegami Image Annotator is seamlessly integrated into Zegami Machine Learning Suite. It turbo charges the curation of your training data. Quickly and visually identify data and image quality issues. Subset your data into classes, automatically cluster similar looking images and identify bias. Having this overview of your data saves days of wasted effort when image quality issues are discovered part way through an annotation project. 

How it works

Once the target images have been identified, you can then view and annotate them in Zegami Image Annotator. The tool makes it easy to draw bounding boxes, polygons and pixel perfect segmentations with both manual and algorithm driven tools. All annotations are automatically synchronised to Zegami so they can be accessed anywhere and by any of your collaborators. The resulting annotations can be exported in a convenient, customisable format that can then be integrated into your machine learning workflow.  

Learn more about Zegami Image Annotator