Manual switch for object detection model to train on new data

Hi, I would like to know if there is a way to have a manual switch when new data is available and we only want to train the model on some data and not all the new data?

Hi Aisha,

There is no “manual” switch available, but you could create one.

By creating another project to upload the new images to, you can use our python package or the REST API to upload images to it for retraining a model if you’d like to separate the two projects. Just note you’ll want to train from your previous weights checkpoints. I do recommend just adding training data to the same project and utilizing transfer learning for active learning.