top of page

Model Training

Our platform offers an infrastructure for training your own model on your own data.

The motivation behind this feature is to train pose-estimation model that will outperform the current one and achieve more accurate results during the keypoint detection phase in our performance assessment tool, which will ultimately lead to a better performance of the system.

The Data Preparation Pipeline

choose your Training Data

Choose the data you want the model to train on

annotate it manually

Annotate your data with the coco-annotator tool

augment and transform

Perform data augmentation and transformation so it can fit as input to the model

start training

Configure your hyperparamters, generate the model and start the training

Annotator.png
Filtered Json.png
Raw Json.png
Raw Images.png
Matlab.png
Seg Images.png
Final Json.png
Training log.png

You bring the data,

we train the model

Our platform allows developers to train a model from scratch with any data they want.

​

All you need to do is to supply the training data for the model and pass it through our data preparation pipeline, and you are good to go.

Data
bottom of page