We simplify the AI Computer Vision lifecycle

Annotation Accuracy

Use test tasks to measure your annotator performance against ground truth tasks : accuracy, overflow, shortage, confusion matrix score, good annotations and bad annotations.

Dataset Creation

Use S3 integration or Browse internet to build your dataset of images.

Team management and statistics

Handle your team of Annotators and quality assurance analysts. Measure their performance with several metrics accumulated over time.

Python SDK & API integration

You can automate use of the platform by leveraging our API and Python SDK. We also offer callback hooks to notify you when your task is completed by an annotator and accepted through the QA process.

Annotation types

The following annotation types are integrated with our editing tools:

• Images: box, line, polygon, point, segmentation, categorization
• LiDAR/RADAR (PCD): segmentation
• Text labeling and Categorization tools can optionally be integrated

Fast Annotation Tools and workflow

• State-of-the-art combination of AI, algorithms and fast workflow are the factors which will enable you to sprint through the annotation process instead of walking.
• Attachment for the next task is preloaded and there is no need to reload the window for the next task. This enables you to run our cloud software with the speed of running it on premises and without the headaches of scaling or maintenance.

Automatic labeling and Train Models in the loop

Use our pre-trained AI models or integrate your own to pre-label your data. Patch up the results manually to make them perfect and save time.

As you label, we can train models in the background based on your dataset. The more you label, the better predictions you are getting. Coming soon.

Build your ML Dataset in 2 minutes

A quick summary of how to build ML Dataset easily

Annotation Service

If you want to avoid the headache of finding your own team, maintaining it and monitoring, we offer our own annotation service for your convinience. Contact us at contactus@trainingset.ai and we can discuss the requirements.

Contour Detection

Segmentation tasks support our proprietary AI models which aim to explore objects or regions more attentive than the surrounding areas on the image. Simply pick the object and it's surrounding area and we will find the object contours and create a segmentated object.

AI Deploy

...

AI Models

TrainingSet.AI goal is to simplify the computer vision life cycle for AI and be able to have production ready AI models without the need of being a data scientist. This is the first in a series of Articles and new features we will be working on.

Train Custom Models

Through the training process, AI should learn to perform its task from labeled data. Head over to the projects view on the dashboard, tune the parameters if you want and then click on the train button and wait.
Feel free to skip the hyperparameter tuning if you want, if your dataset is good, the default parameters will give you solid results as well. If you don’t understand what the hyperparameters do, there is info for each of them.
Time to wait depends on a number of tasks you’ve labeled. You can monitor the log for live updates. Our project took an hour to train with 4284 images.
Read more...

Polygon

Our platform will draw and label polygons around objects on each image/video

Cuboid

Our platform will draw and label cuboids around objects on each image/video

Line and Spline Annotation

Our platform will draw and label lines and splines around objects on each image/video

Semantic Segmentation

Our platform will do a full Semantic Segmentation on each image

Lidar

Our platform will do a full Semantic Segmentation on each Point Cloud Data file

Point Annotation

Our platform will draw and label points around objects on each image/video

Image Classification

Our platform will draw and label lines and splines around objects on each image/video